Not sure the dial-up analogy fits, instead I tend to think we are in the mainframe period of AI, with large centralised computing models that are so big and expensive to host, only a few corporations can afford to do so. We rent a computing timeshare from them (tokens = punch cards).
I look forward to the "personal computing" period, with small models distributed everywhere...
> I look forward to the "personal computing" period, with small models distributed everywhere...
One could argue that this period was just a brief fluke. Personal computers really took off only in the 1990s, web 2.0 happened in the mid-2000s. Now, for the average person, 95%+ of screen time boils down to using the computer as a dumb terminal to access centralized services "in the cloud".
The personal computing era happened partly because, while there were demands for computing, users' connectivity to the internet were poor or limited and so they couldn't just connect to the mainframe. We now have high speed internet access everywhere - I don't know what would drive the equivalent of the era of personal computing this time.
> We now have high speed internet access everywhere
As I travel a ton, I can confidently tell you, that this is still not true at all, and I’m kinda disappointed that the general rule of optimizing for bad reception died.
> the general rule of optimizing for bad reception died.
Yep, and people will look at you like you have two heads when you suggest that perhaps we should take this into account, because it adds both cost and complexity.
But I am sick to the gills of using software - be that on my laptop or my phone - that craps out constantly when I'm on the train, or in one of the many mobile reception black spots in the areas where I live and work, or because my rural broadband has decided to temporarily give up, because the software wasn't built with unreliable connections in mind.
It's not that bleeding difficult to build an app that stores state locally and can sync with a remote service when connectivity is restored, but companies don't want to make the effort because it's perceived to be a niche issue that only affects a small number of people a small proportion of the time and therefore not worth the extra effort and complexity.
Whereas I'd argue that it affects a decent proportion of people on at least a semi-regular basis so is probably worth the investment.
It's always a small crisis what app/book to install on my phone to give me 5-8 hours of reading while on a plane. I found one - Newsify, combine it with YT caching.
Moving services to the cloud unfortunately relieves a lot of the complexity of software development with respect to the menagerie of possible hardware environments.
it of course leads to a crappy user experience if they don't optimize for low bandwidth, but they don't seem to care about that, have you ever checked out how useless your algorithmic Facebook feed is now? Tons of bandwidth, very little information.
It seems like their measure is time on their website equals money in their pocket and baffling you with BS is a great way to achieve that until you never visit again in disgust and frustration.
I don't think the "menagerie of possible hardware environments" excuse holds much water these days. Even web apps still need to accommodate various screen sizes and resolutions and touch vs mouse input.
Native apps need to deal with the variety in software environments (not to say that web apps are entirely insulated from this), across several mobile and desktop operating systems. In the face of that complexity, having to compile for both x86-64 and arm64 is at most a minor nuisance.
I used to work for a company building desktop tools that were distributed to, depending on the tool, on the low end tens of thousands of users, and on the high end, hundreds of thousands. We had one tool that was nominally used by about a million people but, in actuality, the real number of active users each month was more like 300k.
I was at the company for 10 years and I can only remember one issue where we could not reproduce or figure it out on tools that I worked on. There may have been others for other tools/teams, but the number would have been tiny because these things always got talked about.
In my case the guy with the issue - who'd been super-frustrated by it for a year or more - came up to our stand when we were at a conference in the US, introduced himself, and showed me the problem he was having. He then lent me his laptop overnight[0], and I ended up installing Wireshark to see why he was experiencing massive latency on every keystroke, and what might be going on with his network shares. In the end we managed to apply a fix to our code that sidestepped the issue for users with his situation (to this day, he's been the only person - as far as I'm aware - to report this specific problem).
Our tools all ran on Windows, but obviously there were multiple extent versions of both the desktop and server OS that they were run on, different versions of the .NET runtime, at the time everyone had different AV, plus whatever other applications, services, and drivers they might have running. I won't say it was a picnic - we had a support/customer success team, after all - but the vast majority of problems weren't a function of software/OS configuration. These kinds of issues did come up, and they were a pain in the ass, but except in very rare cases - as I've described here - we were always able to find a fix or workaround.
Nowadays, with much better screensharing and remote control options, it would be way easier to deal with these sorts of problems than it was 15 - 20 years ago.
[0] Can't imagine too many organisations being happy with that in 2025.
Have you ever distributed an app on the PC to more than a million people? It might change your view. Browser issues are a different argument and I agree with you 100% there. I really wish people would pull back and hold everyone to consistent standards but they won't.
I work on a local-first app for fun and someone told me I was simply creating problems for myself and I could just be using a server. But I'm in the same boat as you. I regularly don't have good internet and I'm always surprised when people act like an internet connection is a safe assumption. Even every day I go up and down an elevator where I have no internet, I travel regularly, I go to concerts and music festivals, and so on.
I don't even travel that much, and still have trouble. Tethering at the local library or coffee shops is hit or miss, everything slows down during storms, etc.
One problem I've found in my current house is that the connection becomes flakier in heavy rain, presumably due to poor connections between the cabinet and houses. I live in Cardiff which for those unaware is one of Britain's rainiest cities. Fun times.
Access. You cannot use Starlink on a train, flight, inside buildings, etc. Starlink is also not available everywhere: https://starlink.com/map. Also, it’s not feasible to bring that with me a lot of time, for example on my backpack trips; it’s simply too large.
Because of many reasons. It's not practical to have a Starlink antenna with you everywhere. And then yes, cost is a significant factor too - even in the dialup era satellite internet connection was a thing that existed "everywhere", in theory....
Privacy. I absolutely will not ever open my personal files to an LLM over the web, and even with my mid-tier M4 Macbook I’m close to a point where I don’t have to. I wonder how much the cat is out of the back for private companies in this regard. I don’t believe the AI companies founded on stealing IP have stopped.
Not a single person I know that has any apple device would claim that, nobody cares or even knows in detail stuff we discuss here. Its HN bubble at its best.
Another point is, subjectively, added privacy compared to say South Korean products is mostly a myth. It 100% doesn't apply if you are not US citizen and even then, fingers crossed all 3-letter agencies and device creator are not over-analyzing every single data point about you continuously, is naive. What may be better is devices are harder to steal & take ownership, but for that I would need to see some serious independent comparison, not some paid PR from which HN is not completely immune to.
Centralized only became mainstream when everything started to be offered "for free". When it was buy or pay recurrently more often the choice was to buy.
There are no FOSS alternatives for consumer use unless the consumer is an IT pro or a developer. Regular people can’t use most open source software without help. Some of it, like Linux desktop stuff, has a nice enough UI that they can use it casually but they can’t install or configure or fix it.
Making software that is polished and reliable and automatic enough that non computer people can use it is a lot harder than just making software. I’d say it’s usually many times harder.
I don't think that is a software issue but a social issue nowadays. FOSS alternatives have become quite OK in my opinion.
If computers came with Debian, Firefox and Libre Office preinstalled instead of only W11, Edge and with some Office 365 trail, the relative difficulty would be gone I think.
Same thing with most IT departments only dealing with Windows in professional settings. If you even are allowed to use something different you are on your own.
Some people, but a majority see it as free. Go to your local town center and randomly poll people how much they pay for email or google search, 99% will say it is free and stop there.
> We now have high speed internet access everywhere
This is such a HN comment illustrating how little your average HN knows of the world beyond their tech bubble. Internet everywhere, you might have something of a point. But "high speed internet access everywhere" sounds like "I haven't travelled much in my life".
I don't know, I think you're conflating content streaming with central compute.
Also, is percentage of screentime the relevant metric? We moved TV consumption to the PC, does that take away from PCs?
Many apps moved to the web but that's basically just streamed code to be run in a local VM. Is that a dumb terminal? It's not exactly local compute independent...
Nearly entirety of the use cases of computers today don't involve running things on a 'personal computer' in any way.
In fact these days, every one kind of agrees as little as hosting a spreadsheet on your computer is a bad idea. Cloud, where everything is backed up is the way to go.
This. Although briefly, there was at least a couple of years of using pc's without an internet connection. It's unthinkable now. And even back then, when you blinked with your eyes this time period was over.
The personal computer arguably begins with VisiCalc in 1979.
> Through the 1970s, personal computers had proven popular with electronics enthusiasts and hobbyists, however it was unclear why the general public might want to own one. This perception changed in 1979 with the release of VisiCalc from VisiCorp (originally Personal Software), which was the first spreadsheet application.
Mainstream use of the web really took off in the second half of the 1990s. Arbitrarily, let's say with the release of Windows 95. That's a quarter of a century you'd be blinking for.
In time Mainframes of this age will make a come back.
This whole idea that you can connect lots of cheap low capacity boxes and drive down compute costs is already going away.
In time people will go back to thinking compute as a variable of time taken to finish processing. That's the paradigm in the cloud compute world- you are billed for the TIME you use the box. Eventually people will just want to use something bigger that gets things done faster, hence you don't have to rent them for long.
Umm... I had a PC a decade before the web was invented, and I didn't even use the web for like another 5 years after it went public ("it's an interesting bit of tech but it will obviously never replace gopher...")
The killer apps in the 80s were spreadsheets and desktop publishing.
It is true that browsers do much more computation than "dumb" terminals, but there are still non-trivial parallels. Terminals do contain a processor and memory in order to handle settings menus, handle keyboard input and convert incoming sequences into a character array that is then displayed on the screen. A terminal is mostly useless without something attached to the other side, but not _completely_ useless. You can browse the menus, enable local echo, and use device as something like a scratchpad. I once drew up a schematic as ascii art this way. The contents are ephemeral and you have to take a photo of the screen or something in order to retain the data.
Web browsers aren't quite that useless with no internet connection, some sites do offer offline capabilities (for example gmail). but even then, the vast majority of offline experiences exist to tide the user over until network can be re-established, instead of truly offering something useful to do locally. Probably the only mainstream counter-examples would be games.
Google's own Gemma models are runnable locally on a Pixel 9 Max so some lev of AI is replicatable client side. As far as Gmail running locally, it wouldn't be impossible for Gmail to be locally hosted and hit a local cache which syncs with a server only periodically over IMAP/JMAP/whatever if Google actually wanted to do it.
The gain, as far as local AI goes for Google, is that, at Google scale, the CPU/GPU time to run even a small model like Gemma will add up across Gmail's millions of users. If clients have the hardware for it (which Pixel 9's have) it means Gmail's servers aren't burning CPU/GPU time on it.
As far as how Gmail's existing offline mode works, I don't know.
To make the discussion easier, we can look at the state of GMail before widespread AI was a thing.
(I have to use some weaselwording here, because GMail had decent spam detection since basically forever, and whether you call that AI or not depends on where we have shifted the goalposts at the moment.)
Heh that brings to mind that old chestnut, something like "We call it AI until it works, then we call it machine learning." Google's running the servers tho, so the assumption is the servers would be running the spam filter against it before the user. In my mind, anyway. So it's fair to point out that Google also has the compute resources to not have to care. To the hypothetical though, could Google make Gmail a local-first? I say yes. Will they? I doubt it.
Oh, GMail is definitely a cloud offering---even if they have some offline functionality.
I was just probing the 'content _streaming_' term. As you demonstrate, you'd have to squint really hard to describe GMail as content streaming.
'Offline' vs 'content streaming' is a false dichotomy. There's more different types of products and services.
(Which reminds me a bit of crypto-folks calling everything software that's not in crypto "web2", as if working on stodgy backends in a bank or making Nintendo Switch games has anything to do with the web at all.)
Ok sure but there's plenty of simple video streaming in total screen time, which was the context I was replying to. I never claimed it was a dichotomy, simply a large part of screen time that clearly skews the analysis.
Depends on the app, and the personal device. Mobile devices are increasingly thin clients. Of course hardware-wise they are fully capable personal computers, but ridiculous software-imposed limitations make that increasingly difficult.
IPads are incredibly advanced. Though I guess you mean they don't use anything that requires more sophistication from the user (or something like that)?
I'm not questioning whether the Ipad can be used as a client in some capacity, or whether people tend to use it as a client. I question whether the Ipad is a thin client. The answer to that question doesn't lie in how many applications require an internet connection, but in how many applications require local computational resources.
The Ipad is a high performance computer, not just because Apple think that's fun, but out of necessity given its ambition: the applications people use on it require local storage and rather heavy local computation. The web browser standards if nothing else have pretty much guaranteed that the age of thin clients is over: a client needs to supply a significant amount of computational resources and storage to use the web generally. Not even Chromebooks will practically be anything less than rich clients.
Going back to the original topic (and source of the analogy), IOS hosts an on-device large language model.
As with everything, the lines are a bit blurred these days. We may need a new term for these devices. But despite all the compute and storage and on-device models these supercomputers are barely a step above thin clients.
I mean, Chromebooks really aren't very far at all from thin clients. But even my monster ROG laptop when it's not gaming is mostly displaying the results of computation that happened elsewhere
There are more PCs and serious home computing setups today than there were back then. There are just way way way more casual computer users.
The people who only use phones and tablets or only use laptops as dumb terminals are not the people who were buying PCs in the 1980s and 1990s, or they were they were not serious users. They were mostly non-computer-users.
Non-computer-users have become casual consumer level computer users because the tech went mainstream, but there's still a massive serious computer user market. I know many people with home labs or even small cloud installations in their basements, but there are about as many of them as serious PC users with top-end PC setups in the late 1980s.
I dislike the view of individuals as passive sufferers of the preferences of big corporations.
You can and people do self-host stuff that big tech wants pushed into the cloud.
You can have a NAS, a private media player, Home Assistant has been making waves in the home automation sphere. Turns out people don't like buying overpriced devices only to have to pay a $20 subscription, and find out their devices don't talk to each other, upload footage inside of their homes to the cloud, and then get bricked once the company selling them goes under and turns of the servers.
This. And the hordes of people reacting with some explanation for why this is. The 'why' is not the point, we already know the 'why'. The point is that you can if you want. Might not be easy, might not be convenient, but that's not the point. No one has to ask someone else for permission to use other tech than big tech.
The explanation of 'why' is not an argument. Big tech is not making it easy != it's impossible. Passive sufferers indeed.
Edit: got a website with an RSS feed somewhere maybe? I would like to follow more people with a point of view like yours.
I think it has more to do with the 'common wisdom' dictating that this is the way to do it, as 'we've always done it like this'.
Which might even be true, since cloud based software might offer conveniences that local substitutes don't.
However this is not an inherent property of cloud software, its just some effort needs to go into a local alternative.
That's why I mentioned Home Assistant - a couple years ago, smart home stuff was all the rage, and not only was it expensive, the backend ran in the cloud, and you usually paid a subscription for it.
Nowadays, you can buy a local Home Assistant hub (or make one using a Pi) and have all your stuff only connect to a local server.
The same is true for routers, NAS, media sharing and streaming to TV etc. You do need to get technical a bit, but you don't need to do anything you couldn't figure out by following a 20 minute Youtube video.
I look forward to a possibility where the dumb terminal is less centralized in the cloud, and more how it seems to work in the expanse. They all have hand terminals that seem to automatically interact with the systems and networks of the ship/station/building they're in. Linking up with local resources, and likely having default permissions set to restrict weird behavior.
Not sure it could really work like that IRL, but I haven't put a ton of thought into it. It'd make our always-online devices make a little more sense.
But for a broader definition of "personal computer", the number of computers we have has only continued to skyrocket - phones, watches, cars, TVs, smart speakers, toaster ovens, kids' toys...
I'm with GP - I imagine a future when capable AI models become small and cheap enough to run locally in all kinds of contexts.
Depending on how you are defining AI models, they already do. Think of the $15 security camera that can detect people and objects. That is AI model driven. LLM's are another story, but smaller, less effective ones can and do already run at the edge.
Makes me want to unplug and go back to offline social media. That's a joke. The dominant effect was networked applications getting developed, enabling community, not a shift back to client terminals.
Once up on a time social media was called Usenet and worked offline in a dedicated client with a standard protocol. You only went online to download and send messages, but could then go offline and read them in an app of your choice.
Web2.0 discarded the protocol approach and turned your computer into a thin client that does little more than render webapps that require you to be permanently online.
That 'average' is doing a lot of work to obfuscate the landscape. Open source continues to grow (indicating a robust ecosystem of individuals who use their computers for local work) and more importantly, the 'average' looks like it does not necessarily due to a reduction in local use, but to an explosion of users that did not previously exist (mobile first, SAAS customers, etc.)
The thing we do need to be careful about is regulatory capture. We could very well end up with nothing but monolithic centralized systems simply because it's made illegal to distribute, use, and share open models. They hinted quite strongly that they wanted to do this with deepseek.
There may even be a case to be made that at some point in the future, small local models will outperform monoliths - if distributed training becomes cheap enough, or if we find an alternative to backprop that allows models to learn as they infer (like a more developed forward-forward or something like it), we may see models that do better simply because they aren't a large centralized organism behind a walled garden. I'll grant that this is a fairly polyanna take and represents the best possible outcome but it's not outlandishly fantastic - and there is good reason to believe that any system based on a robust decentralized architecture would be more resilient to problems like platform enshittification and overdeveloped censorship.
At the end of the day, it's not important what the 'average' user is doing, so long as there are enough non-average users pushing the ball forward on the important stuff.
The small mind could have an advantage if it is closer or more trustworthy to users.
It only has to be good enough to do what we want. In the extreme, maybe inference becomes cheap enough that we ask “why do I have to wake up the laptop’s antenna?”
I don't want to send sensitive information to a data center, I don't want it to leave my machine/network/what have you. Local models can help in that department.
You could say the same about all self-hosted software, teams with billions of dollars to produce and host SaaS will always have an advantage over smaller, local operations.
Abundant resources could enable bad designs. I could in particular see a lot of commercial drive for huge models that can solve a bazillion different use cases, but aren't efficient for any of them.
There might be also local/global bias strategies. A tiny local model trained on your specific code/document base may be better aligned to match your specific needs than a galaxy scale model. If it only knows about one "User" class, the one in your codebase, it might be less prone to borrowing irrelevant ideas from fifty other systems.
The advantage it might have won't be in the form of "more power", it would be in the form of "not burdened by sponsored content / training or censorship of any kind, and focused on the use-cases most relevant to the individual end user."
We're already very, very close to "smart enough for most stuff". We just need that to also be "tuned for our specific wants and needs".
Even the most popular games (with few exceptions) present as relatively dumb terminals that need constant connectivity to sync every activity to a mainframe - not necessarily because it's an MMO or multiplayer game, but because it's the industry standard way to ensure fairness. And by fairness, of course, I mean the optimization of enforcing "grindiness" as a mechanism to sell lootboxes and premium subscriptions.
And AI just further normalizes the need for connectivity; cloud models are likely to improve faster than local models, for both technical and business reasons. They've got the premium-subscriptions model down. I shudder to think what happens when OpenAI begins hiring/subsuming-the-knowledge-of "revenue optimization analysts" from the AAA gaming world as a way to boost revenue.
But hey, at least you still need humans, at some level, if your paperclip optimizer is told to find ways to get humans to spend money on "a sense of pride and accomplishment." [0]
I imagine there are plenty of indie single-player games that work just fine offline. You lose cloud saves and achievements, but everything else still works.
Funny you would pick this analogy. I feel like we’re back in the mainframe era. A lot of software can’t operate without an internet connection. Even if in practice they execute some of the code on your device, a lot of the data and the heavyweight processing is already happening on the server. Even basic services designed from the ground up to be distributed and local first - like email (“downloading”) - are used in this fashion - like gmail. Maps apps added offline support years after they launched and still cripple the search. Even git has GitHub sitting in the middle and most people don’t or can’t use git any other way. SaaS, Electron, …etc. have brought us back to the mainframe era.
It's always struck me as living in some sort of bizaro world. We now have these super powerful personal computers, both handheld (phones) and laptops (My M4 Pro smokes even some desktop class processors) and yet I use all this powerful compute hardware to...be a dumb terminal to someone else's computer.
I had always hoped we'd do more locally on-device (and with native apps, not running 100 instances of chromium for various electron apps). But, it's hard to extract rent that way I suppose.
I don't even understand why computer and phone manufacturers even try to make their devices faster anymore, since for most computing tasks, the bottleneck is all the data that needs to be transferred to and from the modern version of the mainframe.
There are often activities that do require compute though. My last phone upgrade was so Pokemon Go would work again, my friend upgrades for the latest 4k video or similar.
Yet manufacturers give us thinner and thinner phones every year (instead of using that space for the battery), and make it difficult to swap out batteries which have degraded.
> make it difficult to swap out batteries which have degraded.
That's the part that pisses me off the most. They all claim it's for the IP68, but that's bullshit. There's plenty of devices with removable backs & batteries that are IP68.
My BlackBerry bold 9xxx was 10mm thin. the iPhone 17 Pro Max is 8.75. You aren't going to notice the 1.3mm of difference, and my BlackBerry had a user replaceable battery, no tools required just pop off the back cover.
The BlackBerry was also about 100 grams lighter.
The non-user removable batteries and unibody designs are purely for planned obsolescence, nothing else.
Also when a remote service struggle I can switch to do something else. When a local software struggles it brings my whole device to its knees and I can't do anything.
I think people have been finding more compelling use cases for the fact that information systems can be multi-player now than for marginal FLOPS. Client-server is just a very effective way of organizing multi-player information systems.
> Why would companies sell you the golden goose when they can instead sell you an egg every day?
Because someone else can sell the goose and take your market.
Apple is best aligned to be the disruptor. But I wouldn’t underestimate the Chinese government dumping top-tier open-source models on the internet to take our tech companies down a notch or ten.
By that logic none of us should be paying monthly subscriptions for anything because obviously someone would disrupt that pricing model and take business away from all the tech companies who are charging it? Especially since personal computers and mobile devices get more and more powerful and capable with every passing year. Yet subscriptions also get more prevalent every year.
If Apple does finally come up with a fully on-device AI model that is actually useful, what makes you think they won't gate it behind a $20/mo subscription like they do for everything else?
> By that logic none of us should be paying monthly subscriptions for anything because obviously someone would disrupt that pricing model and take business away from all the tech companies who are charging it?
Non sequitur.
If a market is being ripped off by subscription, there is opportunity in selling the asset. Vice versa: if the asset sellers are ripping off the market, there is opportunity to turn it into a subscription. Business models tend to oscillate between these two for a variety of reasons. Nothing there suggets one mode is infinitely yielding.
> If Apple does finally come up with a fully on-device AI model that is actually useful, what makes you think they won't gate it behind a $20/mo subscription like they do for everything else?
If they can, someone else can, too. They can make plenty of money selling it straight.
Sure, the company that launched iTunes and killed physical media, then released a phone where you can't install apps ("the web is the apps") will be the disruptor to bring back local computing to users...
It's also that same Apple that fights tooth and nail every single attempt to let people have the goose or even the promise of a goose. E.g. by saying that it's entitled to a cut even if a transaction didn't happen through Apple.
It's a lot more complicated than that. They need to be able to take the island very quickly with a decapitation strike, while also keeping TSMC from being sabotaged or destroyed, then they need to be able to weather a long western economic embargo until they can "break the siege" with demand for what they control along with minor good faith concessions.
It's very risky play, and if it doesn't work it leaves China in a much worse place than before, so ideally you don't make the play unless you're already facing some big downside, sort of as a "hail Mary" move. At this point I'm sure they're assuming Trump is glad handing them while preparing for military action, they might even view invasion of Taiwan as defensive if they think military action could be imminent anyhow.
Destroying TSMC or knowing it would be sabotaged would pretty much be the point of the operation. Would take 48 hours and they could be out of there again and say “ooops sorry” at the UN.
Hard disagree. They need chips bad, and it's the US defense position that TSMC be destroyed if possible in the event of successful Chinese invasion. They also care about reunification on principle, and an attack like that without letting them force "One China" on the Taiwanese in the aftermath would just move them farther from that goal.
> then they need to be able to weather a long western economic embargo until they can "break the siege" with demand for what they control along with minor good faith concessions
And you know we'd be potting their transport ships, et cetera, from a distance the whole time, all to terrific fanfare. The Taiwan Strait would become the new training ground for naval drones, with the targets being almost exclusively Chinese.
I worked with the Taiwanese Military, that's their dream scenario but the reality is they're scared shitless that the Chinese will decapitate them with massive air superiority. Drones don't mean shit without C2.
China has between 5:1 and 10:1 advantage depending on asset class. If not already on standby, US interdiction is ~48 hours. For sure China is going to blast on all fronts, so cyber and grid interruptions combined with shock and awe are definitely gonna be a thing. It's not a great setup for Taiwan.
You could say the same thing about Computers when they were mostly mainframe. I am sure someone will figure out how to make it commoditized just like personal computers and internet.
And notably, those phones and tablets are intentionally hobbled by the device owners (Apple, Google) who do everything they can to ensure they can't be treated like personal computing devices. Short of regulatory intervention, I don't see this trend changing anytime soon. We're going full on in the direction of more locked down now that Google is tightening the screws on Android.
They didn't create them, but PC startups like Apple and Commodore only made inroads into the home -- a relatively narrow market compared to business. It took IBM to legitimize PCs as business tools.
> I look forward to the "personal computing" period, with small models distributed everywhere...
Like the web, which worked out great?
Our Internet is largely centralized platforms. Built on technology controlled by trillion dollar titans.
Google somehow got the lion share of browser usage and is now dictating the direction of web tech, including the removal of adblock. The URL bar defaults to Google search, where the top results are paid ads.
Your typical everyday person uses their default, locked down iPhone or Android to consume Google or Apple platform products. They then communicate with their friends over Meta platforms, Reddit, or Discord.
The decentralized web could never outrun money. It's difficult to out-engineer hundreds of thousands of the most talented, most highly paid engineers that are working to create these silos.
Ok, so Brave Browser exists - if you download, you will see 0 ads on the internet, I've never really seen ads on the internet - even in the before brave times.
Fr tho, no ads - I'm not making money off them, I've got no invite code for you, I'm a human - I just don't get it. I've probably told 500 people about Brave, I don't know any that ever tried it.
I don't ever know what to say. You're not wrong, as long as you never try to do something else.
I was gonna say this. If Google decides to stop developing chromium then Brave is left with very few choices.
As someone who has been using brace since it was first announced and very tightly coupled to the BAT crypto token I must say it is much less effective nowadays.
I often still see a load of ads and also regularly have to turn off the shields for some sites.
How?? I watch Hulu ad supported without ads, YouTube no ads ever, screen never goes black, Viki ad supported free package - never seen an ad, Paramount ad package - never seen an ad.
I never have to turn off shields - on one hand the number of times I've had to do that.
Maybe I have something additional installed I don't know.
We have a ton of good, small models. The issues are:
1. Most people don't have machines that can run even midsized local models well
2. The local models are nearly as good as the frontier models for a lot of use cases
3. There are technical hurdles to running local models that will block 99% of people. Even if the steps are: download LM Studio and download a model
Maybe local models will get so good that they cover 99% of normal user use cases and it'll be like using your phone/computer to edit a photo. But you'll still need something to make it automatic enough that regular people use it by default.
That said, anyone reading this is almost certainly technical enough to run a local model. I would highly recommend trying some. Very neat to know it's entirely run from your machine and seeing what it can do. LM Studio is the most brainless way to dip your toes in.
As the hype is dying down it's becoming a little bit clearer that AI isn't like blockchain and might be actually useful (for non generative purposes at least)
I'm curious what counts as a midsize model; 4B, 8B, or something larger/smaller?
What models would you recommend? I have 12GB of vram so anything larger than 8B might be really slow, but i am not sure
Micro: Runs on a microcontroller or GPUs with just 4GB of VRAM
There's really nothing worthwhile for general use cases that runs in under 16GB (from my testing) except a grammar-checking model that I can't remember the name of at the moment.
gpt-oss:20b runs on 16GB of VRAM and it's actually quite good (for coding, at least)! Especially with Python.
Prediction: The day that your average gaming PC comes with 128GB of VRAM is the day developers will stop bothering with cloud-based AI services. gpt-oss:120b is nearly as good as gpt5 and we're still at the beginning of the AI revolution.
Try WebLLM - it's pretty decent and all in-browser/offline even for light tasks, 1B-1.5B models like Qwen2.5-Coder-1.5B-Instruct. I put together a quick prototype - CodexLocal.com but you can essentially a local nginx and use webllm as an offline app. Of course, you can just use Ollama / LM Studio but that would require a more technical solution
I like to think of it more broadly, and that we are currently in the era of the first automobile. [0]
LLMs are the internal combustion engine, and chatbot UIs are at the "horseless carriage" phase.
My personal theory is that even if models stopped making major advancements, we would find cheaper and more useful ways to use them. In the end, our current implementations will look like the automobile pictured below.
I'm not a big google fan, but I really like the "Google AI Edge Gallery" android app [0]. In particular, I've been chatting with the "Gemma-3n-E2B-it" model when I don't have an internet connection, and it's really decent!
The "enshittification" hasn't happened yet. They'll add ads and other gross stuff to the free or cheap tiers. Some will continue to use it, but there will be an opportunity for self-hosted models to emerge.
> Like 50% of internet users are already interacting with one of these daily. You usually only change your habit when something is substantially better.
No, you usually only change your habit when the tools you are already using are changed without consulting you, and the statistics are then used to lie.
I use gpt4all and have downloaded some models. It's not that hard, and doesn't take a huge system. It's pretty easy if you use models on their list, it's a bit more work to find the right chat interactive script (forgot what they are called) for your class of model if you downloaded one that's not on the list, but not that hard. I have 8GB VRAM, and it's slow but they work. I like that it's local and private, but then I don't use them for much other than as an oddity that is fun to interact with.
Mainframes still exist, and they actually make a lot of sense from physics perspective. It's good idea to run transactions in a big machine rather than distributed, the latter is less energy efficient.
I think the misconception is that things cannot be overpriced for reasons other than inefficiency.
this -- chips are getting fast enough both arm n x86. unified memory architecture means we can get more ram on devices at faster throughput. we're already seeing local models - just that their capability is limited by ram.
I actually don’t look forward to this period. I have always been for open source software and distributism — until AI.
Because if there’s one thing worse than governments having nuclear weapons, it’s everyone having them.
It would be chaos. And with physical drones and robots coming, it woukd be even worse. Think “shitcoins and memecoins” but unlike those, you don’t just lose the money you put in and you can’t opt out. They’d affect everyone, and you can never escape the chaos ever again. They’d be posting around the whole Internet (including here, YouTube deepfakes, extortion, annoyance, constantly trying to rewrite history, get published, reputational destruction at scale etc etc), and constant armies of bots fighting. A dark forest.
And if AI can pay for its own propagation via decentralized hosting and inference, then the chance of a runaway advanced persistent threat compounds. It just takes a few bad apples, or even practical jokers, to unleash crazy stuff. And it will never be shut down, just build and build like some kind of kessler syndrome. And I’m talking about with just CURRENT AI agent and drone technology.
Funny how this guy thinks he knows exactly what's up with AI, and how "others" are "partly right and wrong." Takes a bit of hubris to be so confident. I certainly don't have the hubris to think I know exactly how it's all going to go down.
How about a vague prediction that covers all scenarios? XD
*ahem* It's gonna be like every other tool/societal paradigm shift like the smartphone before this, and planes/trains/cars/ships/factories/electricity/oil/steam/iron/bronze etc. before that:
• It'll coalesce into the hands of a few corporations.
• Idiots in governments won't know what the fuck to do with it.
• Lazy/loud civvies will get lazier/louder through it.
• There'll be some pockets of individual creativity and freedom, like open source projects, that will take varying amounts of time to catch on in popularity or fade away to obscurity.
• One or two killer apps that seem obvious but nobody thought of, will come out of nowhere from some nobody.
• Some groups will be quietly working away using it to enable the next shift, whether they know it or not.
• Aliens will land turning everything upside down. (I didn't say when)
The problem is that the bubble people are so unimaginative, similar to Krugman, that those who have any inkling of an imagination can literally feel like visionaries compared to them. I know I’m describing Dunning-Krueger, but so be it, the bubble people are very very wrong. It’s like, man, they really are unable to imagine a very real future.
It’s a weird comparison since internet in the dial-up age was a bubble, are you saying the hype machine for AI is in fact smaller than the internet? Are you implying that AI will in fact grow that much more slowly and sustainably than the internet, despite trillions of investment?
Do you think Sam Altman, Jeff Bezos, and Mark Zuckerberg are all wrong saying that we’re in a bubble? Do they “lack imagination?”
Also? What do I need imagination for, isn’t that what AI does now?
Not only was it a bubble but we actually had the internet. We didn't have to use our imaginations and build for the technology yet to be invented and we still had a massive damaging bubble.
Even beyond that, the Soviet Union had just collapsed. The US had a balanced budget and 10X less debt. Globalization was just getting started. China was nothing compared to what it was today. It was the absolute most stable time of my life. Basically the opposite of 2025.
"This time things are different because AI blah blah blah".
When you boil it down, all bubbles are based around the idea of using your imagination to image how "this time it is different".
The people who don't think this is a bubble are not seeing that their imagination is a bug and not a feature here.
I find the argument for the bubble to be extremely straightforward.
Currently, investment into AI exceeds the dot-com bubble by a factor of 17. Even in the dot-com era, the early internet was already changing media and commerce in fundamental ways. November is the three-year anniversary of ChatGPT. How much economic value are they actually creating? How many people are purchasing AI-generated goods? How much are people paying for AI-provided services? The value created here would have to exceed what the internet was generating in 2000 by a factor of 17 (which seems excessive to me) to even reach parity with the dot-com bubble.
"But think where it'll be in 5 years"—sure, and let's extrapolate that based on where it is now compared to where it was 3 years ago. New models present diminishing returns. 3.5 was groudbreaking; 4 was a big step forward; 5 is incremental. I won't deny that LLMs are useful, and they are certainly much more productized now than they were 3 years ago. But the magnitude of our collective investment in AI requires that a huge watershed moment be just around the corner, and that makes no sense. The watershed moment was 3 years ago. The first LLMs created a huge amount of potential. Now we're realizing those gains, and we're seeing some real value, but things are also tapering off.
Surely we will have another big breakthrough some day—a further era of AI which brings us closer to something like AGI—but there's just no reason to assume AGI will crop up in 2027, and nothing less that that can produce the ROI that such enormous valuations will eventually, inexorably, demand.
I don’t get why people find it so hard to understand that a technology can be value-additive and still be in a position of massive overinvestment. Every generation of Californians seeks to relive the 1848 gold rush, spending millions excavating rivulets for mere ounces of (very real!) gold.
What is AGI in your mind? Let's take someone who once upon a time was responsible for grading papers. As far as that person is concerned, AGI has arrived for their profession (it arrived nearly two years ago for them). You'll never be better than something that has read every book ever and can write better than you. AGI will come in tranches. Are you really going to hire that developer because you need extra manpower to stand up test coverage? No, so as far as that developer is concerned, AGI has arrived for that part of their professional career.
The bet is not that there will be this one seminal moment of AGI where all the investment will make sense. The bet is that it has already showed up if you look for specific things and will continue to do so. I wouldn't bet against the idea that LLMs will introduce itself to all jobs, one at a time. Reddit moderators, for example, will meet AGI (as far as they know, their entire world being moderating) sooner than say, I don't know, a Radiologist.
The universe of people getting paid to make CRUD apps is over. Many here will be introduced to AGI faster and sooner. Then it could be phone customer support representatives. It could show up for the face-to-face worker who is now replaced by a screen that can talk to customers (which already arrived yesterday, it's here). It'll appear erratic and not cohesive, unless you zoom out and see the contagion.
---
Rome needed to recognize that the Barbarian hordes had arrived. Pay attention to all the places the invasion has landed. You can pretend like the Vandals are not in your town for a little bit, sure, but eventually they will be knocking on many doors (most likely all doors). We're in a time period of RADICAL transformation. There is no half-assing this conviction. Practicality will not serve us here.
> Are you really going to hire that developer because you need extra manpower to stand up test coverage? No, so as far as that developer is concerned, AGI has arrived for that part of their professional career.
That is exactly what you need in order to make AI useful. Even a baby needs to cry to signal its needs to parents, which are like ASI to it. AI working on a task lacks in 3 domains: start, middle and finish.
AI cannot create its own needs, they belong to the context where it is used. After we set AI to work, it cannot predict the outcomes of its actions unless they pass through your context and return as feedback. In the end, all benefits accumulate in the same context. Not to mention costs and risks - they belong to the context.
The AI is a generalist, context is exactly what it lacks. And context is distributed across people, teams, companies. Context is non-fungible. You can't eat so I get satiated. Context is what drives AI. And testing is the core contextual activity when using AI.
> Let's take someone who once upon a time was responsible for grading papers. As far as that person is concerned, AGI has arrived for their profession
You're talking about TAs. I know TAs. Their jobs have not disappeared. They are not using AI to grade papers.
> Are you really going to hire that developer because you need extra manpower to stand up test coverage?
Yes. Unsupervised AI agents cannot currently replace developers. "Oh we'll get someone to supervise it"—yes, that person's job title is "developer" and they will be doing largely the same job they'd have done 5 years ago.
> The universe of people getting paid to make CRUD apps is over.
Tell that to all the people who get paid to make CRUD apps. Frankly, Airtable has done more to disrupt CRUD apps than AI ever did.
> Rome needed to recognize that the Barbarian hordes had arrived.
IDK what to tell you. All these jobs are still around. You're just fantasizing.
Can you seriously deny that it will take many fewer developers using LLMs to do what was done without LLMs? Many if not most software companies have either greatly slowed hiring, or laid off tranches of developers.
Not all developer jobs will disappear, but there will most certainly be fewer available. Any new grad can tell you how hard it is to find a software engineering job nowadays.
Certainly we cannot just let an AI spin and build software unattended. But what used to take days can now be done in minutes.
> How many people are purchasing AI-generated goods?
Probably a lot. I remember my mom recently showing me an AI-generated book she bought. And pretty much immediately refunded it. Not because it was AI, but because the content was trash.
> Even in the dot-com era, the early internet was already changing media and commerce in fundamental ways.
I agree that AI is overhyped but so was the early web. It was projected to do a lot of things ”soon”, but was not really doing that much 4 years in. I don’t think the newspapers or commerce were really worried about it. The transformation of the business landscape took hold after the crash.
This is not true. Obviously the underlying effect is real but not nearly to this scale—for instance, neither the CPI nor the S&P500 are even remotely close to 17x higher than they were at the turn of the millennium.
He figured there was a credit bubble like that around the time of the dot com bubble and now but the calculation if purely based on interest rates and the money can go into any assets - property, stocks, crypto etc. It's not AI specific.
> Robots combined with AI will be creating that value.
Will they? Within what timeframe? Because a bubble economy can't be told to "just hang on a few more years" forever. LLMs are normal technology; they will not suddenly become something they are not. There's no indication that general intelligence is right on the horizon.
If you think they wont then you should short the stocks starting with Nvidia and get rich.
> There's no indication that general intelligence is right on the horizon.
You dont need general intelligence for all the tasks, if a robot can do some of those tasks with limited intelligence cheaper than a humnan, that is all corporations care about.
Almost everyone I hear calling our AI hype machine a bubble aren't claiming AI is a short term fluke. They're saying the marketing doesn't match the reality. The companies don't have the revenue they need. The model performance is hitting the top of the S curve. Essentially, this is the first big wave - but it'll be a while before the sea level rises permanently.
It’s not just a marketing stunt, it’s a trillion dollar grift that VCs are going to try to dump off onto the public markets when the reality doesn’t catch up to the hype fast enough
The LLM architectures we have now have reached their full potential already, so going further would require something completely different. It isn’t a matter of refining the existing tech, whereas the internet of 1997 is virtually technologically identical to what we have today. The real change has been sociological, not technological.
To make a car analogy; the current LLMs are not the early cars, but the most refined horse drawn carriages. No matter how much money is poured into them, you won’t find the future there.
The current generation of LLM's have convinced me that we already have the compute and the data needed for AGI, we just likely need a new architecture. But I really think such an architecture could be right around the corner. It appears to me like the building blocks are there for it, it would just take someone with the right luck and genius to make it happen.
> The current generation of LLM's have convinced me that we already have the compute and the data needed for AGI, we just likely need a new architecture.
I think this is one of the greatest fallacies surrounding LLMs. This one, and the other one - scaling compute!! The models are plenty fine, what they need is not better models, or more compute, they need better data, or better feedback to keep iterating until they reach the solution.
Take AlphaZero for example, it was a simple convolutional network, not great compared to LLMs, small relative recent models, and yet it beat the best of us at our own game. Why? Because it had unlimited environment access to play games against other variants of itself.
Same for the whole Alpha* family, AlphaStar, AlphaTensor, AlphaCode, AlphaGeometry and so on, trained with copious amounts of interactive feedback could reach top human level or surpass humans in specific domains.
What AI needs is feedback, environments, tools, real world interaction that exposes the limitations in the model and provides immediate help to overcome them. Not unlike human engineers and scientists - take their labs and experiments away and they can't discover shit.
It's also called the ideation-validation loop. AI can ideate, it needs validation from outside. That is why I insist the models are not the bottleneck.
For Alpha Zero, the "better data" was trivial. The environment of board games is extremely simplistic. It just can't be compared to language models.
The problem with language is that there is no know correct answer. Everything is vague, ambiguous and open ended. How would we even implement feedback for that?
> The current generation of LLM's have convinced me that we already have the compute and the data needed for AGI, we just likely need a new architecture
This is likely true but not for the reasons you think about. This was arguably true 10 years ago too. A human brain uses 100 watts per day approx and unlike most models out there, the brain is ALWAYS in training mode. It has about 2 petabytes of storage.
In terms of raw capabilities, we have been there for a very long time.
The real challenge is finding the point where we can build something that is AGI level with the stuff we have. Because right now, we might have the compute and data needed for AGI but we might lack the tools needed to build a system that efficient. It's like a little dog trying to enter a fenced house, the closest path topologically between the dog and the house might not be accessible for that dog at that point because its current capabilities (short legs, inability to jump high or push through the fence standing in between) so while it is further topologically, a longer path topologically might be the closest path to reach the house.
In case it's not obvious, AGI is the house, we are the little dog and the fence represent current challenges to build AGI.
The notion that the brain uses less energy than an incandescent lightbulb and can store less data than YouTube does not mean we have had the compute and data needed to make AGI "for a very long time".
The human brain is not a 20-watt computer ("100 watts per day" is not right) that learns from scratch on 2 petabytes of data. State manipulations performed in the brain can be more efficient than what we do in silicon. More importantly, its internal workings are the result of billions of years of evolution, and continue to change over the course of our lives. The learning a human does over its lifetime is assisted greatly by the reality of the physical body and the ability to interact with the real world to the extent that our body allows. Even then, we do not learn from scratch. We go through a curriculum that has been refined over millennia, building on knowledge and skills that were cultivated by our ancestors.
An upper bound of compute needed to develop AGI that we can take from the human brain is not 20 watts and 2 petabytes of data, it is 4 billion years of evolution in a big and complex environment at molecular-level fidelity. Finding a tighter upper bound is left as an exercise for the reader.
> it is 4 billion years of evolution in a big and complex environment at molecular-level fidelity. Finding a tighter upper bound is left as an exercise for the reader.
You have great points there and I agree. Only issue I take with your remark above. Surely, by your own definition, this is not true. Evolution by natural selection is not a deterministic process so 4 billion years is just one of many possible periods of time needed but not necessarily the longest or the shortest.
Also, re "The human brain is not a 20-watt computer ("100 watts per day" is not right)", I was merely saying that there exist an intelligence that consumes 20 watts per day. So it is possible to run an intelligence on that much energy per day. This and the compute bit do not refer to the training costs but to the running costs after all, it will be useless to hit AGI if we do not have enough energy or compute to run it for longer than half a millisecond or the means to increase the running time.
Obviously, the path to design and train AGI is going to take much more than that just like the human brain did but given that the path to the emergence of the human brain wasn't the most efficient given the inherent randomness in evolution natural selection there is no need to pretend that all the circumstances around the development of the human brain apply to us as our process isn't random at all nor is it parallel at a global scale.
> Evolution by natural selection is not a deterministic process so 4 billion years is just one of many possible periods of time needed but not necessarily the longest or the shortest.
That's why I say that is an upper bound - we know that it _has_ happened under those circumstances, so the minimum time needed is not more than that. If we reran the simulation it could indeed very well be much faster.
I agree that 20 watts can be enough to support intelligence and if we can figure out how to get there, it will take us much less time than a billion years. I also think that on the compute side for developing the AGI we should count all the PhD brains churning away at it right now :)
Dial-up modems reached their full 56kbps potential in 1997, and going further required something completely different. It happened naturally to satisfy demand, and was done by many of the same companies and people; the change was technological, not sociological.
I think we're probably still far from the full potential of LLMs, but I don't see any obstacles to developing and switching to something better.
I don't think that comparison works very well at all.
We had plenty of options for better technologies both available and in planning, 56k modems were just the cost effective/lowest common denominator of their era.
It's not nearly as clear that we have some sort of proven, workable ideas for where to go beyond LLMs.
> Dial-up modems reached their full 56kbps potential in 1997
That's simply not true. Modems were basically the same tech in the dsl era, and using light instead of electricity is a very gradual refinement.
> we're probably still far from the full potential of LLMs
Then how come the returns are so extremely diminishing?
> I don't see any obstacles to developing and switching to something better.
The obstacle is that it needs to be invented. There was nothing stopping newton from discovering relativity either. We simply have no idea what the road forward even looks like.
what we can say right now is that we've hit the point of diminishing returns and the only way we're going to get signicantly more capable models is through a technological advance that we cannot forsee (and that may not come for decades if it ever comes)
You could see some potential modifications. Already some are multimodal. You'd probably want something to change the weights as time goes on so they can learn. It might be more steam engines needing to be converted to petrol engines.
> If you told someone in 1995 that within 25 years [...] most people would find that hard to believe.
That's not how I remember it (but I was just a kid so I might be misremembering?)
As I remember (and what I gather from media from the era) late 80s/early 90s were hyper optimistic about tech. So much so that I distinctly remember a ¿german? TV show when I was a kid where they had what amounts to modern smartphones, and we all assumed that was right around the corner. If anything, it took too damn long.
Were adults outside my household not as optimistic about tech progress?
To your point, AT&T's "You Will" commercials started airing in 1993 and present both an optimistic and fairly accurate view of what the future would look like.
I didn't know about these ads, thanks for sharing! Can't imagine how people reacted to that when they aired — the things they described sound so "normal" today, I wonder if it was seen as far fetched, crazy or actually expected.
I was in my late teens at the time. My memory is that I felt like the tech was definitely going happen in some form, but I rolled my eyes heavily at the idea that AT&T was going to be the company to do make it happen.
If you’re unfamiliar, the phone connectivity situation in the 80s and 90s was messy and piecemeal. AT&T had been broken up in 1982 (see https://www.historyfactory.com/insights/this-month-in-busine...), and most people had a local phone provider and AT&T was the default long-distance provider. MCI and Sprint were becoming real competition for AT&T at the time of these commercials.
Anyway, in 1993 AT&T was still the crusty old monopoly on most people’s minds, and the idea that they were going to be the company to bring any of these ideas to the market was laughable. So the commercials were basically an image play. The only thing most people bought from AT&T was long distance service, and the main threat was customers leaving for MCI and Sprint. The ads memorable for sure, but I don’t think they blew anyone’s mind or made anyone stay with AT&T.
We’re the same age, and I had exactly the same reaction.
AT&T and the baby bells were widely loathed (man I hated Ameritech…), so the idea they would extend their tentacles in this way was the main thing I reacted to. The technology seemed straightforwardly likely with Dennard scaling in full swing.
I thought it would be banks that owned the customer relationship, not telcos or Apple (or non-existent Google), but the tech was just… assume miniaturization’s plateau isn’t coming for a few decades.
In these commercials, it wasn't the technology itself but the ease of access and visualized integration of these technologies into the commoners' everyday lives that was the new idea.
Or hell, as Neil deGrasse Tyson said in a video, just put 2 lines with different arrows at the ends then our brains can't even tell if they're the same size!
That’s how I remember it too. The video is from 1999, during the height of the dot-com bubble. These experts are predicting that within 10 years the internet will be on your phone, and that people will be using their phones as credit cards and the phone company would manage the transaction, the prediction actually comes pretty close to the prediction made by bitcoin enthusiasts.
To be fair, that has been a Sci-Fi trope for at least 130 years and predates the invention of the car itself (e.g. personal wings/flying horse -> flying ship -> personal balloon -> flying automobile). So countless generations have been waiting for that :)
There's no way I'm trusting the current driving cohort with a third dimension. If we get flying cars and they aren't completely autonomous, I am moving to the sticks.
The vast majority of the dot-com comparison that I personally see are economic, not technological. People (or at least the ones I see) are claiming that the bubble mechanics of e.g. circular trading and over-investments are similar to the dot-com bubble, not that the AI technology is somehow similar the internet (it obviously isn’t). And to that extent we are in the year 1999 not 1995.
When this article are claiming both sides of the debate, I believe only one of them are real (the ones hyping up the technology). While there are people like me who are pessimistic about the technology, we are not in any position of power, and our opinion on the matter is basically a side noise. I think a much more common (among people with any say in the future of this technology) is the believe that this technology is not yet at a point which warrants all this investment. There were people who said that about the internet in 1999, and they were proven 100% correct in the months that followed.
I'm getting ai fatigue. It's ok to rewrite quick emails that i'm having brain farts on but anything deep it just sucks. I certainly can't see paying for it.
Weird because AI has been solving hard problems for me. Even finding solutions that I couldn’t find myself. Ie. sometimes my brain cant wrap around a problem, I throw it to AI and it perfectly solves it.
It is weird that AI is solving hard problems for you. I can't get it to do the most basic things consistently, most of the time it's just pure garbage. I'd never pay for "AI" because it wastes more of my time than it saves. But I've never had a problem wrapping my head around a problem, I solve problems.
I'm curious what kind of problem your "brain cant wrap around", but the AI could.
I'm curious what kind of problem your "brain cant wrap around", but the AI could.
One of the most common use cases is that I can't figure out why my SQL statement is erroring or doesn't work the way it should. I throw it into ChatGPT and it usually solves it instantly.
Yes. To me, it is. Sometimes queries I give it are 100-200 lines long. Sure, I can solve it eventually but getting an "instant" answer that is usually correct? Absolutely priceless.
It's pretty common for me to spend a day being stuck on a gnarly problem in the past. Most developers have. Now I'd say that's extremely rare. Either an LLM will solve it outright quickly or I get enough clues from an LLM to solve it efficiently.
You might be robbing yourself of the opportunity to learn SQL for real by short-cutting to a solution that might not even be correct one.
I've tried using LLMs for SQL and it fails at exactly that: complexity. Sure it'll get the basic queries right, but throw in anything that's not standard every day SQL into it and it'll give you solutions that are not great really confidently.
If you don't know SQL enough to figure out these issues in the first place, you don't know if the solutions the LLM provides are actually good or not. That's a real bad place to be in.
Have you ever read Zen and the Art of Motorcycle Maintenance? One of the first examples in that book is how when you are disassembling a motorcycle any one bolt is trivial until one is stuck. Then it becomes your entire world for a while as you try to solve this problem and the solution can range from trivial to amazingly complex.
You are using the term “hard problem” to mean something like solving P = NP. But in reality as soon as you step outside of your area of expertise most problems will be hard for you. I will give you some examples of things you might find to be hard problems (without knowing your background):
- what is the correct way to frame a door into a structural exterior wall of a house with 10 foot ceilings that minimized heat transfer and is code compliant.
- what is the correct torque spec and sequence for a Briggs and Stratton single cylinder 500 cc motor.
- how to correctly identify a vintage Stanley hand plane (there were nearly two dozen generations of them, some with a dozen different types), and how to compare them and assess their value.
- how to repair a cracked piece of structural plastic. This one was really interesting for me because I came up with about 5 approaches and tried two of them before asking an LLM and it quickly explained to me why none of the solutions I came up with would work with that specific type of plastic (HDPE is not something you can glue with most types of resins or epoxies and it turns out plastic welding is the main and best solution). What it came up with was more cost efficient, easier, and quicker than anything I thought up.
- explaining why mixing felt, rust, and CA glue caused an exothermal reaction.
- find obscure local programs designed to financially help first time home buyers and analyze their eligibility criteria.
In all cases I was able to verify the solutions. In all cases I was not an expert on the subject and in all cases for me these problems presented serious difficulty so you might colloquially refer to them as hard problems.
In this case, the original author stated that AI only good for rewriting emails. I showed a much harder problem that AI is able to help me with. So clearly, my problem can be reasonably described as “hard” relative to rewriting emails.
What happens when these "AI" companies start charging you what it really costs to run the "AI"? You'd very likely balk at it and have to learn SQL yourself. Enjoy it while it lasts, I guess?
I work with some very complex queries (that I didn't write), and yeah, AI is an absolute lifesaver, especially in troubleshooting situations. What used to take me hours now takes me minutes.
In my case, Learning new stuff is one place I see AI playing major role. Especially the academic research which is hard to start if you are newbie but with AI I can start my research, read more papers with better clarity.
Calculate the return on investment for a solar installation of a specified size on a specified property based on the current dynamic prices of all of the panels, batteries, inverter, and balance of system components, the current zoning and electrical code, the current cost of capital, the average insolation and weather taking into account likely changes in weather in the future as weather instability increases due to more global increase of temperature, the chosen installation method and angle, and the optimal angle of the solar panels if adjusted monthly or quarterly. Now do a Manual J calculation to determine the correct size of heat pump in each section of that property, taking into account number of occupants, insulation level, etc.
ChatGPT is currently the best solar calculator on the publicly accessible internet and it's not even close. It'll give you the internal rate of return, it'll ask all the relevant questions, find you all the discounts you can take in taxes and incentives, determine whether you should pay the additional permitting and inspection cost for net metering or just go local usage with batteries, size the batteries for you, and find some candidate electricians to do the actual installation once you acquire the equipment.
Edit: My guess is that it'd cost several thousand dollars to hire someone to do this for you, and it'll save you probably in the $10k-$30k range on the final outcomes, depending on the size of system.
My God, the first example is having an AI do math, then he says "Well I trust it to a standard deviation"
So it's literally the same as googling "what's the ballpark solar installation cost for X in Y area" unbelievable, and people pay $20+ per month for this
Where they agree it shows the data supports that answer - not necessarily that it is true, where they disagree it shows you need to hedge. That's useful.
I would recommend spending that "couple thousand" for quote(s). It's a second opinion from someone who hopefully has high volume in your local market. And your downside could be the entire system plus remediation, fines, etc.
To be clear, I'm not opposed to experimenting, but I wouldn't rely on this. Appreciate your comment for the discussion.
No I'm not relying on it in the sense of going out and running the entire project through it, but as an accurate screener for whether it's worth doing, there's nothing comparable available.
Well deep/hard is different I guess; I use it, day and night, for things I find boring. Boilerplate coding (which now is basically everything that's not pure business logic / logic / etc), corporate docs, reports etc. Everything I don't want to do is done by AI now. It's great. Outside work I use it for absolutely nothing though; I am writing a book, framework and database; that's all manual work (and I don't AI is good at any of those (yet)).
As an LLM-skeptic who got a Claude subscription, the free models are both much dumber and configured for low latency and short dumb replies.
No it won’t replace my job this year or the next, but what Sonnet 4.5 and GPT 5 can do compared to e.g. Gemini Flash 2.5 is incredible. They for sure have their limits and do hallucinate quite a bit once the context they are holding gets messy enough but with careful guidance and context resets you can get some very serious work done with them.
I will give you an example of what it can’t do and what it can: I am working on a complicated financial library in Python that requires understanding nuanced parts of tax law. Best in class LLM cannot correctly write the library code because the core algorithm is just not intuitive. But it can:
1. Update all invocations of the library when I add non-optional parameters that in most cases have static values. This includes updating over 100 lengthy automated tests.
2. Refactor the library to be more streamlined and robust to use. In my case I was using dataclasses as the base interface into and out of it and it helped me split one set of classes into three: input, intermediate, and output while fully preserving functionality. This was a pattern it suggested after a changing requirement made the original interface not make nearly as much sense.
3. Point me to where the root cause of failing unit tests was after I changed the code.
4. Suggest and implement a suite of new automated tests (though its performance tests were useless enough for me to toss out in the end).
5. Create a mock external API for me to use based on available documentation from a vendor so I could work against something while the vendor contract is being negotiated.
6. Create comprehensive documentation on library use with examples of edge cases based on code and comments in the code. Also generate solid docstrings for every function and method where I didn’t have one.
7. Research thorny edge cases and compare my solutions to commercial ones.
8. Act as a rubber ducky when I had to make architectural decisions to help me choose the best option.
It did all of the above without errors or hallucinations. And it’s not that I am incapable of doing any of it, but it would have taken me longer and would have tested my patience when it comes to most of it. Manipulating boilerplate or documenting the semantic meaning between a dozen new parameters that control edge case behavior only relevant to very specific situations is not my favorite thing to do but an LLM does a great job of it.
I do wish LLMs were better than they are because for as much as the above worked well for me, I have also seen it do some really dumb stuff. But they already are way too good compared to what they should be able to do. Here is a short list of other things I had tried with them that isn’t code related that has worked incredibly well:
- explaining pop culture phenomenon. For example I had never understood why Dr Who fans take a goofy campy show aimed in my opinion at 12 year olds as seriously as if it was War and Peace. An LLM let me ask all the dumb questions I had about it in a way that explained it well.
- have a theological discussion on the problem of good and evil as well as the underpinnings of Christian and Judaic mythology.
- analyze in depth my music tastes in rock and roll and help fill in the gaps in terms of its evolution. It actually helped me identify why I like the music I like despite my tastes spanning a ton of genres, and specifically when it comes to rock, created one of the best and most well curated playlists I had ever seen. This is high praise for me since I pride myself on creating really good thematic playlists.
- help answer my questions about woodworking and vintage tool identification and restoration. This stuff would have taken ages to research on forums and the answers would still be filled with purism and biased opinions. The LLM was able to cut through the bullshit with some clever prompting (asking it to act as two competing master craftsmen).
- act as a writing critic. I occasionally like to write essays on random subjects. I would never trust an LLM to write an original essay for me but I do trust it to tell me when I am using repetitive language, when pacing and transitions are off, and crucially how to improve my writing style to take it from B level college student to what I consider to be close to professional writer in a variety of styles.
Again I want to emphasize that I am still very much on the side of there being a marketing and investment bubble and that what LLMs can do being way overhyped. But at the same time over the last few months I have been able to do all of the above just out of curiosity (the first coding example aside). These are things I would have never had the time or energy to get into otherwise.
You seem very thoughtful and careful about all this, but I wonder how you feel about the emergence of these abilities in just 3 years of development? What do you anticipate it will be capable of in the next 3 years?
With no disrespect I think you are about 6-12 months behind SOTA here, the majority of recent advances have come from long running task horizons. I would recommend to you try some kind of IDE integration or CLI tool, it feels a bit unnatural at first but once you adapt your style a bit, it is transformational. A lot of context sticking issues get solved on their own.
Oh I am very much catching up. I am suing Claude Code primarily, and also have been playing a bit with all the latest API goodies from OpenAI and Anthropic, like custom tools, memory use, creating my own continuous compaction algorithm for a specific workflow I tried. There is a lot happening here very fast.
One thing that struck me: models are all trained starting 1-2 years ago. I think the training cutoff for Sonnet 4.5 is like May 2024. So I can only imagine with is being trained and tested currently. And also these models are just so ahead of things like Qwen and Llama for the types of semi-complex non-coding tasks I have tried (like interpreting my calendar events), that it isn’t even close.
“But the fact that some geniuses were laughed at does not imply that all who are laughed at are geniuses. They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown.”
Because some notable people dismissed things that wound up having profound effect on the world, it does not mean that everything dismissed will have a profound effect.
We could just as easily be "peak Laserdisc" as "dial-up internet".
I was happy to come into this thread and see I was not the first person for whom that quote came to mind. The dial-up Internet comparison implicitly argues for a particular outcome of current AI as a technology, but doesn't actually support that argument.
There's another presumably unintended aspect of the comparison that seems worth considering. The Internet in 2025 is certainly vastly more successful and impactful than the Internet in the mid-90s. But dial-up itself as a technology for accessing the Internet was as much of a dead-end as Laserdisc was for watching movies at home.
Whether or not AI has a similar trajectory as the Internet is separate from the question of whether the current implementation has an actual future. It seems reasonable to me that in the future we're enjoying the benefits of AI while laughing thinking back to the 2025 approach of just throwing more GPUs at the problem in the same way we look back now and get a chuckle out of the idea of "shotgun modems" as the future.
While I mostly agree with the article's premise (that AI will cause more software development to happen, not less) I disagree with two parts:
1. the opening premise comparing AI to dial-up internet; basically everyone knew the internet would be revolutionary long before 1995. Being able to talk to people halfway across the world on a BBS? Sending a message to your family on the other side of the country and them receiving it instantly? Yeah, it was pretty obvious this was transformative. The Krugman quote is an extreme, notable outlier, and it gets thrown out around literally every new technology, from blockchain to VR headsets to 3DTVs, so just like, don't use it please.
2. the closing thesis of
> Consider the restaurant owner from earlier who uses AI to create custom inventory software that is useful only for them. They won’t call themselves a software engineer.
The idea that restaurant owners will be writing inventory software might make sense if the only challenge of creating custom inventory software, or any custom software, was writing the code... but it isn't. Software projects don't fail because people didn't write enough code.
Before I got my first full time software engineering gig (I had worked part time briefly years prior) I was working full time as a carpenter. We were paying for an expensive online work order system. Having some previous experience writing software for music in college and a couple /brief/ LAMP stack freelance jobs after college I decided to try to write my own work order system. It took me like a month and it would never have never scaled, was really ugly, and had the absolute minimum number of features. I could never had accepted money from someone to use it but it did what we needed and we ran with it for several years after that.
I was only able to do this because I had some prior programming experience but I would imagine that if AI coding tools get a bit better they would enable a larger cohort of people to build a personal tool like I did.
I don't think his quote is that extreme and it was definitely not obvious to most people. A common thing you heard even around 95 was "I've tried internet but it was nothing special".
> basically everyone knew the internet would be revolutionary long before 1995. Being able to talk to people halfway across the world on a BBS? Sending a message to your family on the other side of the country and them receiving it instantly? Yeah, it was pretty obvious this was transformative.
That sounds pretty similar to long-distance phone calls? (which I'm sure was transformative in its own way, but not on nearly the same scale as the internet)
Do we actually know how transformative the general population of 1995 thought the internet would or wouldn't be?
In 1995 in France we had the minitel already (like really a lot of people had one) and it was pretty incredible, but we were longing for something prettier, cheaper, snappier and more point to point (like the chat apps or emails).
As soon as the internet arrived, a bit late for us (I'd say 1999 maybe) due to the minitel's "good enough" nature, it just became instantly obvious, everyone wanted it. The general population was raving mad to get an email address, I never heard anyone criticize the internet like I criticize the fake "AI" stuff now.
>> Consider the restaurant owner from earlier who uses AI to create custom inventory software that is useful only for them. They won’t call themselves a software engineer.
I have a suspicion this is LLM text, sounds corny. There are dozens open source solutions, just look one up.
Great analysis but one thing overlooked is that current gen advanced AI could in five or ten years (or less) be run from the smartphone or desktop, which could negate all the capex from the hyperscalers and also Nvidia, which presents a massive target for competitors right now. The self same AI revolution we’re seeing created right now could take itself down if AI tooling becomes widespread.
It's more like the Segway era when people with huge stakes in Segway tried to convince the world we were about to rebuild entire cities around the new model.
I recall the unit economics making sense for all these other industries and bubbles (short of maybe tulips, which you could plant…) . Sure there were over-valuation bubbles because of speculatory demand, but right now the assumption seems to be “first to AGI wins” but that… may not happen.
The key variable for me in this house of cards is how long folks will wait before they need to see their money again, and whether these companies will go in the right direction long enough given these valuations to get to AGI. Not guaranteed and in the meantime society will need to play ball (also not a guarantee)
The article seems well researched, has some good data, and is generally interesting. It's completely irrelevant to the reality of the situation we are currently in with LLMs.
It's falling into the trap of assuming we're going to get to the science fiction abilities of AI with the current software architectures, and within a few years, as long as enough money is thrown at the problem.
All I can say for certain is that all the previous financial instruments that have been jumped on to drive economic growth have eventually crashed. The dot com bubble, credit instruments leading to the global financial crisis, the crypto boom, the current housing markets.
The current investments around AI that we're all agog at are just another large scale instrument for wealth generation. It's not about the technology. Just like VR and BioTech wasn't about the technology.
That isn't to say the technology outcomes aren't useful and amazing, they are just independant of the money. Yes, there are Trillions (a number so large I can't quite comprehend it to be honest) being focused into AI. No, that doesn't mean we will get incomprehensible advancements out the other end.
AGI isn't happening this round folks. Can hallucinations even be solved this round? Trillions of dollars to stop computers lying to us. Most people where I work don't even realise hallucinations are a thing. How about a Trillion dollars so Karen or John stop dismissing different viewpoints because a chat bot says something contradictory, and actually listen? Now that would be worth a Trillion dollars.
Imagine a world where people could listen to others outside of their bubble. Instead they're being given tools that re-inforce the bubble.
People keep comparing the AI boom to the Dotcom bubble. They’re wrong. Others point to the Railway Mania of the 1840s — closer, but still not quite right.
The real parallel is Canal Mania — Britain’s late-18th-century frenzy to dig waterways everywhere. Investors thought canals were the future of transport. They were, but only briefly.
Today’s AI runs on GPUs — chips built for rendering video games, not thinking machines. Adapting them for AI is about as sensible as adapting a boat to travel across land. Sure, it moves — but not quickly, not cheaply, and certainly not far.
It works for now, but the economics are brutal. Each new model devours exponentially more power, silicon, and capital. It just doesn't scale.
The real revolution will come with new, hardware built for the job (that hasn't been invented yet) — thousands of times faster and more efficient. When that happens, today’s GPU farms will look like quaint relics of an awkward, transitional age: grand, expensive, and obsolete almost overnight.
I think specialized hardware will emerge for specific proven workloads (transformer inference, for example), but GPUs won't become obsolete. They'll remain the experimentation platform for new architectures. You need flexibility to discover what's worth building custom silicon for.
Think 3D printers versus injection molds: you prototype with flexibility, then mass-produce with purpose-built tooling. We've seen this pattern before too. CPUs didn't vanish when GPUs arrived for graphics. The canal analogy assumes wholesale replacement. Reality is likely more boring: specialization emerges and flexibility survives.
> Today’s AI runs on GPUs — chips built for rendering video games, not thinking machines. Adapting them for AI is about as sensible as adapting a boat to travel across land.
A GPU is fundamentally just a chip for matrix operations, and that's good for graphics but also for "thinking machines" as we currently have them. I don't think it's like a boat traveling on land at all.
Another definition: A modern GPU a general purpose computer that can make parallelized and efficient computations. It's optimized to run limited number of operations but on large number of data points.
This happens to be useful both for graphics (same "program" running on on huge number of pixels/vertices) and neural networks (same neural operations on huge number of inputs/activations)
GPUs are massively parallel, sure, but they still have a terrible memory architecture and are difficult to program (and are still massively memory constrained). It's only NVidia's development in cuda that made it even feasible to create decent ML models on GPUs.
I think it'll be a combination of hardware of course, but also better software - surely there is a better way of doing this (like our brains do) which will eventually require less power
Recently, in my city, the garbage trucks started to come equipped with a device I call "The Claw" (think Toy Story). The truck drives to your curb where your bin is waiting, and then The Claw extends, grasps the bin, lifts it into the air and empties the contents into the truck before setting it down again.
The Claw allows a garbage truck to be crewed by one man where it would have needed two or three before, and to collect garbage much faster than when the bins were emptied by hand. We don't know what the economics of such automation of (physical) garbage collection portend in the long term, but what we do know is that sanitation workers are being put out of work. "Just upskill," you might say, but until Claw-equipped trucks started appearing on the streets there was no need to upskill, and now that they're here the displaced sanitation workers may be in jeopardy of being unable to afford to feed their families, let alone find and train in some new marketable skill.
So no, we're in the The Claw era of AI, when business finds a new way to funge labor with capital, devaluing certain kinds of labor to zero with no way out for those who traded in such labor. The long-term implications of this development are unclear, but the short-term ones are: more money for the owner class, and some people are out on their ass without a safety net because this is Goddamn America and we don't brook that sort of commie nonsense here.
FYI, this kind of garbage truck has been around for >50 years [0], so any wide-scale impact on employment from this technology has likely already settled out.
The waste collection companies in my area don't use them because it's rural and the bins aren't standardized. The side loaders don't work for all use cases of garbage trucks.
>In 1969, the city of Scottsdale, Arizona introduced the world's first automated side loader. The new truck could collect 300 gallon containers in 30 second cycles, without the driver exiting the cab
I feel like I've been in such a phase for a good 10-15 years now. I miss looking back and marvel at the technological progress we've made, and eagerly anticipating the next exciting thing. For a while now, I look back wondering why so few people seem bothered by the fact that technology becomes worse every year, with fewer features, lower quality, and more stupidity; and new developments are no longer about exciting improvements, but dreading what new thing that used to work perfectly fine is broken now.
For a bit, I thought science and industry were finally starting to see the problem with our quality degradation and tech regression. Instead, the current hype cycle is all about settling for even crappier quality and lower reliability.
how much does the correction here hew to making an AI model just look like standardized API calls with predictable responses? If you took away all the costs (data centers, water consumption, money, etc) I still wouldn't use an LLM as a first choice because it's wrong enough of the time to make it useless -- I have to verify everything it says, which is how I would have approached a task in the first place. If we put that analogy into manufacturing, it's "I have to QA everything off of the line _without exception_ and I get frequent material waste"
If you make the context small enough, we're back at /api/create /api/read /api/update /api/delete; or, if you're old-school, a basic function
The only problem is, similarity with dotcom might only go thus far. For example, dotcom bubble itself might not have a similar thing in the past at that time. This is because the overall world context is different and interaction of social, political and economic forces is different.
So, when people say something about future, they are looking into the past to draw some projections or similar trends, but they may be missing the change in the full context. The considered factors of demand and automation might be too few to understand the implications. What about political, social and economic landscape? The systems are not so much insulated to study using just a few factors.
In the dial-up era, the industry was young, there were no established players, it was all a big green field.
The situation is far from similar now. Now there's an app for everything and you must use all of them to function, which is both great and horrible.
From my experience, current generation of AI is unreliable and so cannot be trusted. It makes non-obvious mistakes and often sends you off on tangents, which consumes energy and leads to confusion.
It's an opinion I've built up over time from using AI extensively.
I would have expected my opinion to improve after 3 years, but it hasn't.
People tend to equate this to the railroad boom when saying that infrastructure spending will yield durable returns into the future no matter what.
When the railroad bubble popped we had railroads. Metal and sticks, and probably more importantly, rights-of-way.
If this is a bubble, and it pops, basically all the money will have been spent on Nvidia GPUs that depreciate to 0 over 4 years. All this GPU spending will need to be done again, every 4 years.
Hopefully we at least get some nuclear power plants out of this.
Yeah, the short-lived GPU deprecation cycle does feel very relevant here.
I'm still a fan of the railroad comparisons though for a few additional reasons:
1. The environmental impact of the railroad buildout was almost incomprehensibly large (though back in the 1800s people weren't really thinking about that at all.)
2. A lot of people lost their shirts investing in railroads! There were several bubbly crashes. A huge amount of money was thrown away.
3. There was plenty of wasted effort too. It was common for competing railroads to build out rails that served the same route within miles of each other. One of them might go bust and that infrastructure would be wasted.
Imagine the progress we could have made on climate change if this money had been funneled into that, instead of making some GPU manufacturers obscenely wealthy.
What percentage of data centre build costs are the GPUs vs power stations, water cooling plants, buildings, roads, network, racks, batteries, power systems, etc
A bunch of the money is being spent on data centers and their associated cooling and power systems and on the power plants and infrastructure. Those should have much longer depreciation schedules.
There's a lot more to infrastructure spending than GPUs. Companies are building data centers, cooling systems, power plants (including nuclear), laying cables under oceans, launching satellites. Bubble or not, all of this will continue to be useful for decades in the future.
Heck if nothing else all the new capacity being created today may translate to ~zero cost storage, CPU/GPU compute and networking available to startups in the future if the bubble bursts, and that itself may lead to a new software revolution. Just think of how many good ideas are held back today because deploying them at scale is too expensive.
>> basically all the money will have been spent on Nvidia GPUs that depreciate to 0 over 4 years
I agree the depreciation schedule always seems like a real risk to the whole financial assumptions these companies/investors make, but a question I've wondered:
- Will there be an unexpected opportunity when all these "useless" GPUs are put out to pasture? It just seems like saying a factory will be useless because nobody wants to buy an IBM mainframe, but an innovative company can repurpose a non-zero part of that infrastructure for another use case.
Railroads need repair too? Not sure if it's every 4 years. Also, the trains I take to/from work are super slow because there is no money to upgrade.
I think we may not upgrade every 4 years, but instead upgrade when the AI models are not meeting our needs AND we have the funding & political will to do the upgrade.
Perhaps the singularity is just a sigmoid with the top of the curve being the level of capex the economy can withstand.
For what it's worth they cost a lot less than highways to maintain. Something like the 101 in the Bay Area costs about $40,000 per lane-mile per year, or about $240,000.
Trains are closer to $50-100,000 per mile per year.
If there's no money for the work it's a prioritization decision.
I think the hardware infrastructure may be obsolete but at the moment we are still just beginning to figure out how to use AI. So the knowledge will be the important thing that’s left after the bubble. The current infrastructure will probably be as obsolete as dial up infrastructure.
This is precisely why the AI bubble is so much worse than previous bubbles: the main capital asset that the bubble is acquiring is going to depreciate before the bubble's participants can ever turn a profit. Regardless of what AI's future capabilities are going to be, it's physically impossible for any of these companies to become profitable before the GPUs that they have already purchased are either obsolete or burnt out from running under heavy load.
There are some gross approximations in the comparison. Oversized fibre optics networks laid out in the late 90s were used for years and may even be in part still used today; today's servers and GPUs will be obsolete in 3 to 5 years, and not worth their weight in scrap metal in 10.
The part about Jevons' paradox is interesting though.
One thing the analysis for textiles vs cars misses I the complexity of the supply chain and the raw materials / components that need to be procured to make the end product. Steel/textiles have simple supply chains and they went through a boom/bust cycle as the demand plateaued. But cars on the other hand will not go through the same pattern - there are too many logistical things that need to line up and the trust factor in each of those steps as well as the end product is quite high.
Software is similar to cars - the individual components that need to be properly procured and put together is very complex and trust will be important - will you trust that you as a restaurant owner vibe coded your payment stack properly or will you just drop in the 3 lines to integrate with Stripe? I think most of the non-tech business owners will do the latter.
It's clear that AI is useful. It's not yet clear how useful. Hype has always obscured real value, and nobody knows the real value until the hype cycle completes.
What is clear, is that we have strapped a rocket to our asses, fueled with cash and speculation. The rocket is going so fast we don't know where we're going to land, or if we'll land softly, or in a very large crater. The past few decades have examples of craters. Where there are potential profits, there are people who don't mind crashing the economy to get them.
I don't understand why we're allowing this rocket to begin with. Why do we need to be moving this quickly and dangerously? Why do we need to spend trillions of dollars overnight? Why do we need to invest half the fucking stock market on this brand new technology as fast as we can? Why can't we develop it in a way that isn't insanely fast and dangerous? Or are we incapable of decisions not based on greed and FOMO?
Who is "we" ? I certainly don't spend trillions on frivolities. I think the Saudis via Softbank do, and these people build fake cities in the desert, they are by definition dumb money.
They earn so much from oil and are so keenly aware this will stop, they'd rather spend a trillion on a failure, than keep that cash rotting away with no future investment.
No project, no country, can swallow the Saudi oil money like Sam Altman can. So, they're building enormous data centers with custom nuclear plants and call that Stargate to syphon that dumb money in. It's the whole business model of Softbank: find a founder whose hubris is as big as Saudi stupidity.
> The other claims that AI will create more jobs than it destroys.
Maybe it's my bubble, but so far I didn't hear someone saying that. What kind of jobs should that be, given that both forms, physical and knowledge work, will be automatable sooner or later?
That claim just reads like he's concocted two sides for his position to be the middle ground between. I did that essays in high school but I try to be better than that now.
What makes this analogy great is that nobody in the dial up days could imagine Google or YouTube. We’re in the same place now nobody knows who becomes “the Google of AI,” and that uncertainty usually means a new platform is being born.
I would go so far as to say we are still in the computing dial-up era. We're at the tail end, maybe - we don't write machine code any longe, mostly, and we've abstracted up a few levels but we're still writing code. Eventually computing is something that will be everywhere, like air, and natural language interfaces will be nearly exclusively how people interact with computing machines. I don't think the idea of 'writing software' is something that will stick around, I think we're in a very weird and very brief little epoch where that is a thing.
It took a long long time going from a walking bike to the one we know now. It's not going to be different from AI. Transformers will only get us so far and for the rest we need another tock. AGI is not going to happen with this generation of hardware. We are hitting spatial scaling limits in video and image generation and we are hitting limits with LLMs.
The thing is that the average person now thinks AI is revolutionary.
Thus, if you form the analogy correctly, then it tells us that the average person is
wrong and that AI is NOT revolutionary. (I'm not arguing either case.)
There is one key way in which I believe the current AI bubble differs from the TMT bubble. As the author points out, much of the TMT bubble money was spent building infrastructure that benefited us many decades later.
But in the case of AI, that argument is much harder to make. The cost of compute hardware is astronomic relative to the pace of improvements. In other words, a million dollars of compute today will be technically obsolete (or surpassed on a performance/watt basis) much faster than the fiber optic cables laid by Global Crossing.
And the AI data centers specialized for Nvidia hardware today may not necessarily work with the Nvidia (or other) hardware five years from now—at least not without major, costly retrofits.
Arguably, any long-term power generation capacity put down for data centers of today would benefit data centers of tomorrow, but I'm not sure much such investment is really being made. There's talk of this and that project, but my hunch and impression is that much of it will end up being small-scale local power generation from gas turbines and the like, which is harmful for the local environment and would be quickly dismantled if the data center builders or operators hit the skids. In other words, if the bubble bursts I can't imagine who would be first in line to buy a half-built AI data center.
This leads me to believe this bubble has generated much less useful value to benefit us in future than the TMT bubble. The inference capacity we build today is too expensive and ages too fast. So the fall will be that much more painful for the hyperscalers.
> Couldn't AI like be their custom inventory software?
Absolutely not. It's inherently a software with a non-zero amount of probability in every operation. You'd have a similar experience asking an intern to remember your inventory.
Like I enjoy Copilot as a research tool right but at the same time, ANYTHING that involves delving into our chat history is often wrong. I own three vehicles, for example, and it cannot for it's very life remember the year, make and model of them. Like they're there, but they're constantly getting switched around in the buffer. And once I started positing questions about friend's vehicles that only got worse.
But you should be able to say "remember this well" and AI would know it needs a reliable database instead of relying on its LLM cache or whatever. Could it not just spin up Postgres in some Codex Cloud like a human developer would? Not today but in a few years?
Wouldn't an intelligent computer know to use tools? The core of the point being discussed seems to be why do you need to ask it to make it you inventory software when an intelligent system would know that when asked to build an inventory system setting up a database and logging all the information is need and ask agents to do that.
It's the same question that you might have asked in 1920. "This radio hardly works at all. Can't they do something about all the static? I don't see the big deal. This is just a scam to sell batteries and tubes."
A radio isn't intelligent and isn't marketed as such. If you're going to sell software you call intelligent, I don't think I'm out of pocket for saying "this feels even dumber than regular software I use."
(Shrug) Well, I do. I don't know where you people get the idea that these sorts of things spring forth fully-formed from the brow of Zeus, but it doesn't work that way.
Sounds like evasion to me. At some point you'll need to explain exactly why systems that thrash Go grandmasters, make entire forums fail the Turing test, and take gold at international math and programming competitions are "fundamentally not intelligent."
The only argument you (and Toucan, who's been around longer) seem to muster is that the systems aren't perfect. They occasionally say stupid things, need extra handholding, babble nonsense and write buggy code, commit blatant plagiarism, fall into fallacious reasoning traps, and can be fooled with simple tricks... unlike people, presumably.
Limited tests in specific settings are always surprising and do speak to it's use, but are not a testament to it being intelligent.
I replied because you answered "you're doing it wrong" to a question of it's failures. It seems you dismiss the concerns of the smaller errors or failures without realizing the point being made. If it's "smart" enough to solve take international math medals and beat grand masters at go, but can't truly understand problems and anticipate needs or issues, to me it's not a genuine intelligence and in it's current form never will be.
It's not that they are not perfect, is that they have no concept of reality, and it's evident in their failures. Beyond this point I am not interested in trying to convince you.
Great comment threads here, but the OP article leans too much on AI generated text that is heavy on empty synthetic rhetoric and generative AI clichés.
Nice article, but somewhat overstates how bad 1995 was meant to be.
A single image generally took nothing like a minute. Most people had moved to 28.8K modems that would deliver an acceptable large image in 10-20 seconds. Mind you, the full-screen resolution was typically 800x600 and color was an 8-bit palette… so much less data to move.
Moreover, thanks to “progressive jpeg”, you got to see the full picture in blocky form within a second or two.
And of course, with pages was less busy and tracking cookies still a thing of the future, you could get enough of a news site up to start reading in less time that it can take today.
One final irk is that it’s little overdone to claim that “For the first time in history, you can exchange letters with someone across the world in seconds”. Telex had been around for decades, and faxes, taking 10-20 seconds per page were already commonplace.
> Regardless of which specific companies survive, this infrastructure being built now will create the foundation for our AI future - from inference capacity to the power generation needed to support it.
Does that comparison with the fiber infra from the dotcom era really hold up? Even when those companies went broke, the fiber was still perfectly fine a decade later. In contrast, all those datacenters will be useless when the technology has advanced by just a few years.
Nobody is going to be interested in those machines 10 years from now, no matter if the bubble bursts or not. Data centers are like fresh produce. They are only good for a short period of time and useless soon after. They are being constantly replaced.
> from inference capacity to the power generation needed to support it.
Power gen = yes. Reasoning is trivial due to long useful asset life.
For compute, why would it be any different than fiber in the AI bubble pop case? Less useful, but not useless. Are you misremembering the fiber overbuild era? The fiber investments were impaired and sold off for pennies on the dollar. The value here wouldn't go to zero, either.
Really tired of seeing the story about how, “sure Worldcom et al went bankrupt but their investments in fiber optics gave us the physical infrastructure of the Internet today.”
I mean, sort of, but the fiber optics in the ground have been upgraded several by orders of magnitude of its original capacity by replacing the transceivers on either end. And the fiber itself has lasted and will continue to last for decades.
Neither of those properties is true of the current datacenter/GPU boom. The datacenter buildings may last a few decades but the computers and GPUs inside will not and they cannot be easily amplified in their value as the fiber in the ground was.
Mass production of telephone line modems in the United States began as part of the SAGE air-defense system in 1958, connecting terminals at various airbases, radar sites, and command-and-control centers to the SAGE director centers scattered around the United States and Canada.
Shortly afterwards in 1959, the technology in the SAGE modems was made available commercially as the Bell 101, which provided 110 bit/s speeds. Bell called this and several other early modems "datasets".
> 1. Economic strain (investment as a share of GDP)
> 2. Industry strain (capex to revenue ratios)
> 3. Revenue growth trajectories (doubling time)
> 4. Valuation heat (price-to-earnings multiples)
> 5. Funding quality (the resilience of capital sources)
> His analysis shows that AI remains in a demand-led boom rather than a bubble, but if two of the five gauges head into red, we will be in bubble territory.
This seems like a more quantitative approach than most of "the sky is falling", "bubble time!", "circular money!" etc analyses commonly found on HN and in the news. Are there other worthwhile macro-economic indicators to look at?
It's fascinating how challenging it is meaningfully compare current recent events to prior economic cycles such as the y2k tech bubble. It seems like it should be easy but AFAICT it barely even rhymes.
Besides your chart, another point along these lines is that the article cites Azhar claiming multiples are not in bubble territory while also mentioning Murati getting essentially infinite price multiple. Hmmmm...
I feel like this article is too cute. The internet, and the state of the art of computing in general has been driven by one thing and one thing alone: Moore’s Law. In that very real sense, it means that the semiconductor and perhaps more generally even just TSMC is responsible for the rise of the internet and the success of it.
We’re at the end of Moore’s Law, it’s pretty reasonable to assume. 3nm M5 chips means there are—what—a few hundred silicon atoms per transistor? We’re an order of magnitude away from .2 nm which is the diameter of a single silicon atom.
My point is, 30 years have passed since dial up. That’s a lot of time to have exponentially increasing returns.
There’s a lot of implicit assumption that “it’s just possible” to have a Moore’s Law for the very concept of intelligence. I think that’s kinda silly.
Moore's law has very little to do with the physical size of a single transistor. It postulates that the speed and capability of computers will double every few years. Miniaturization is one way to get that increase, but there are other ways.
>The internet, and the state of the art of computing in general has been driven by one thing and one thing alone: Moore’s Law.
You're wrong here... the one thing driving the internet and start of the art computing is money. Period. It wouldn't matter if Moore never existed, and his law was never a thing, money would still be driving technology to improve.
> The one thing driving the internet and state of the art computing is money
You're kind of separating yin from yang and pretending that one begot the other. The reason so much money flooded into chip fab was because compute is one of the few technologies (the only technology?) with recursive self improvement properties. Smaller chip fab leads to more compute, which enabled smaller chip fabs though research modeling. Sure: and it's all because humans want to do business faster. But TSMC literally made chips the business and proved out the pure play foundry business model.
> Even if Moore's Law was never a thing
Then arguably in that universe, we would have eventually hit a ceiling, which is precisely the point I'm trying to make against the article: it's a little silly to assume there's an infinite frontier of exponential improvement available just because that was the prior trend.
> Moore's Law has very little to do with the physical size of a single transistor
I mean it has everything to do with the physical size of a single transistor, precisely because of that recursive self improvement phenomenon. In a universe where moore's law doesn't exist, in 2025 we wouldn't be on 3nm production dies, and compute scale would have capped off decades ago. Or perhaps even a lot of other weird physical things would probably be different, like maybe macroscopic quantum phenomena or just an entire universe that is one sentient blob made from the chemical composition of cheeto dust.
Transistor size is not the only metric that matters in computer speed. Maybe you weren't around when 1MHz CPUs were considered fast. Then there were 8Mhz, then 16MHz, then 25MHz, and soon enough it was 250Mhz, then it jumped up to 1GHz, and now we're seeing 4GHz and faster. We're probably not at the end of the GHz that can be achieved. Chip dies got bigger, too. Way bigger. It doesn't matter if a single transistor can't be shrunk smaller than 3nm if the chip size can be increased. We've seen this in Cerebras Wafer Scale Engine (WSE), which is 12 inches by 12 inches and contains 4 trillion transistors. And then there's the possibility of 3D chip design - if you can't go wider, build taller - but the main problem with all of this is heat and power. More transistors, more GHz, larger dies, all means more heat - and heat is the real limiting factor. If heat and power weren't a concern then we'd have far faster computers.
But all of these advancements in processing power are driven by money, not by some made-up "law" that sounds nice on paper but has little to do with the real world. Sorry but "Moore's law" isn't really a "law" in any way like the laws of physics.
You’ve completely ignored my arguments, you’re hung up on one technicality, and now you’re just being derisive. I literally have a degree in computer engineering. I’m well aware there’s more than just semiconductor size. I’m aware of 3D chip fabs. I’m well aware of clock speed as a dimension. I’m also well fucking aware that moore’s law is not a physical law.
My whole fucking point is that neither are the AI scaling laws.
Good point. Experience teaches us that the breakthroughs in AI will lead to something, we just don’t know what that something is yet and there is a lot of (maybe too much) speculative betting on what it could be.
> most people hit a wall figuring out what it's actually good for beyond parlor tricks.
That's what my parents thought about computers and the internet, wondering what it's actually good for beyond burning $9000 in phone bills to Zerg rush Protoss noobs.
And all the other things computers+internet could do, they could already do through other more reliable (at the time) ways.
But then it turned out that simply making mundane tasks just a little bit faster, and reducing the need to interact with strangers by just that little bit, created a new step on the staircase, a new baseline, with which to reach and do other grander things more easily.
Being able to plan a trip from a single sentence would be one killer app for many people:
"I'm free next week. I'd like to go to A, B, or C for a couple days. What's a cheap flight and a room within this budget near X area?"
and if it could go and also make a booking through your accounts that would be amazing.
But right now even Google's Gemini is an utter useless dumbass if asked to search Google Flights or Airbnb.
I mean if LLMs could just be a natural-language wrapper around existing tools, that'd be amazing in itself. But corporivalry has made that an stillborn dream.
I felt the first half of my reply was too elaborate for such a self-promoting, inaccurate, Nostradamus-knows-better kind of post. I thought it was only sensible that I meet nonsense with nonsense in my closing. Looks like you enjoyed that.
It's not true. Even though the vast majority of a tech company's cost were engineers, they also all grew like crazy and made much more (and grew even more) than they paid in salaries. The OP is correct, if it was just about salaries. These companies would have grown far more and hired far more endlessly, but the complexity of software at that scale makes diminishing returns from hiring more.
If anything the push to use AI has made me more expensive as a dev so far. If im honest i am only slightly more productive but i also burn through tokens.
Most of the big services seem to waste so much time clunking through updating and editing files.
I'm no expert but I can't help feeling there's lots of things they could be doing vastly better in this regard - presumably there is lots to do and they will get around to it.
My head canon is that the thing that preemptively pops the bubble is Apple coming out and saying, very publicly, that AI is a dead end, and they are dropping it completely (no more half assed implicit promises).
And not just that, they come out with an iPhone that has _no_ camera as an attempt to really distance themselves from all the negative press tech (software and internet in particular) has at the moment.
Keyboards were replaced with a touch screen alternative that effectively does the same job though. What is the alternative to a camera? Cameras are way too useful on a mobile device for anyone to even consider dropping them IMO.
That would require people that know about AI to actually choose to cancel it - which nobody that actually knows what AI can do, would ever actually do.
The Apple engineers, with their top level unfettered access to the best Apple AI - they'll convince shareholders to fund it forever, even if normal people never catch on.
Not sure the dial-up analogy fits, instead I tend to think we are in the mainframe period of AI, with large centralised computing models that are so big and expensive to host, only a few corporations can afford to do so. We rent a computing timeshare from them (tokens = punch cards).
I look forward to the "personal computing" period, with small models distributed everywhere...
> I look forward to the "personal computing" period, with small models distributed everywhere...
One could argue that this period was just a brief fluke. Personal computers really took off only in the 1990s, web 2.0 happened in the mid-2000s. Now, for the average person, 95%+ of screen time boils down to using the computer as a dumb terminal to access centralized services "in the cloud".
The personal computing era happened partly because, while there were demands for computing, users' connectivity to the internet were poor or limited and so they couldn't just connect to the mainframe. We now have high speed internet access everywhere - I don't know what would drive the equivalent of the era of personal computing this time.
> We now have high speed internet access everywhere
As I travel a ton, I can confidently tell you, that this is still not true at all, and I’m kinda disappointed that the general rule of optimizing for bad reception died.
> the general rule of optimizing for bad reception died.
Yep, and people will look at you like you have two heads when you suggest that perhaps we should take this into account, because it adds both cost and complexity.
But I am sick to the gills of using software - be that on my laptop or my phone - that craps out constantly when I'm on the train, or in one of the many mobile reception black spots in the areas where I live and work, or because my rural broadband has decided to temporarily give up, because the software wasn't built with unreliable connections in mind.
It's not that bleeding difficult to build an app that stores state locally and can sync with a remote service when connectivity is restored, but companies don't want to make the effort because it's perceived to be a niche issue that only affects a small number of people a small proportion of the time and therefore not worth the extra effort and complexity.
Whereas I'd argue that it affects a decent proportion of people on at least a semi-regular basis so is probably worth the investment.
We ignore the fallacies of distributed computing at our peril: https://en.wikipedia.org/wiki/Fallacies_of_distributed_compu...
It's always a small crisis what app/book to install on my phone to give me 5-8 hours of reading while on a plane. I found one - Newsify, combine it with YT caching.
Usually it reduces not adds complexity. Simpler pages without hundred different js frameworks are faster.
Moving services to the cloud unfortunately relieves a lot of the complexity of software development with respect to the menagerie of possible hardware environments.
it of course leads to a crappy user experience if they don't optimize for low bandwidth, but they don't seem to care about that, have you ever checked out how useless your algorithmic Facebook feed is now? Tons of bandwidth, very little information.
It seems like their measure is time on their website equals money in their pocket and baffling you with BS is a great way to achieve that until you never visit again in disgust and frustration.
I don't think the "menagerie of possible hardware environments" excuse holds much water these days. Even web apps still need to accommodate various screen sizes and resolutions and touch vs mouse input.
Native apps need to deal with the variety in software environments (not to say that web apps are entirely insulated from this), across several mobile and desktop operating systems. In the face of that complexity, having to compile for both x86-64 and arm64 is at most a minor nuisance.
I don't know that it ever held that much water.
I used to work for a company building desktop tools that were distributed to, depending on the tool, on the low end tens of thousands of users, and on the high end, hundreds of thousands. We had one tool that was nominally used by about a million people but, in actuality, the real number of active users each month was more like 300k.
I was at the company for 10 years and I can only remember one issue where we could not reproduce or figure it out on tools that I worked on. There may have been others for other tools/teams, but the number would have been tiny because these things always got talked about.
In my case the guy with the issue - who'd been super-frustrated by it for a year or more - came up to our stand when we were at a conference in the US, introduced himself, and showed me the problem he was having. He then lent me his laptop overnight[0], and I ended up installing Wireshark to see why he was experiencing massive latency on every keystroke, and what might be going on with his network shares. In the end we managed to apply a fix to our code that sidestepped the issue for users with his situation (to this day, he's been the only person - as far as I'm aware - to report this specific problem).
Our tools all ran on Windows, but obviously there were multiple extent versions of both the desktop and server OS that they were run on, different versions of the .NET runtime, at the time everyone had different AV, plus whatever other applications, services, and drivers they might have running. I won't say it was a picnic - we had a support/customer success team, after all - but the vast majority of problems weren't a function of software/OS configuration. These kinds of issues did come up, and they were a pain in the ass, but except in very rare cases - as I've described here - we were always able to find a fix or workaround.
Nowadays, with much better screensharing and remote control options, it would be way easier to deal with these sorts of problems than it was 15 - 20 years ago.
[0] Can't imagine too many organisations being happy with that in 2025.
Have you ever distributed an app on the PC to more than a million people? It might change your view. Browser issues are a different argument and I agree with you 100% there. I really wish people would pull back and hold everyone to consistent standards but they won't.
I work on a local-first app for fun and someone told me I was simply creating problems for myself and I could just be using a server. But I'm in the same boat as you. I regularly don't have good internet and I'm always surprised when people act like an internet connection is a safe assumption. Even every day I go up and down an elevator where I have no internet, I travel regularly, I go to concerts and music festivals, and so on.
I don't even travel that much, and still have trouble. Tethering at the local library or coffee shops is hit or miss, everything slows down during storms, etc.
> everything slows down during storms
One problem I've found in my current house is that the connection becomes flakier in heavy rain, presumably due to poor connections between the cabinet and houses. I live in Cardiff which for those unaware is one of Britain's rainiest cities. Fun times.
Yeah British trains are often absolutely awful for this, I started putting music on my phone locally to deal with the abysmal coverage.
Not true because of cost or access? If you consider starlink high speed, it truly is available everywhere.
Access. You cannot use Starlink on a train, flight, inside buildings, etc. Starlink is also not available everywhere: https://starlink.com/map. Also, it’s not feasible to bring that with me a lot of time, for example on my backpack trips; it’s simply too large.
Because of many reasons. It's not practical to have a Starlink antenna with you everywhere. And then yes, cost is a significant factor too - even in the dialup era satellite internet connection was a thing that existed "everywhere", in theory....
Privacy. I absolutely will not ever open my personal files to an LLM over the web, and even with my mid-tier M4 Macbook I’m close to a point where I don’t have to. I wonder how much the cat is out of the back for private companies in this regard. I don’t believe the AI companies founded on stealing IP have stopped.
Privacy is a niche concern sadly.
I believe Apple has made a significant number of iPhone sales due to a perception of better privacy than Android.
I believe you could be in a bubble.
Not a single person I know that has any apple device would claim that, nobody cares or even knows in detail stuff we discuss here. Its HN bubble at its best.
Another point is, subjectively, added privacy compared to say South Korean products is mostly a myth. It 100% doesn't apply if you are not US citizen and even then, fingers crossed all 3-letter agencies and device creator are not over-analyzing every single data point about you continuously, is naive. What may be better is devices are harder to steal & take ownership, but for that I would need to see some serious independent comparison, not some paid PR from which HN is not completely immune to.
> I don't know what would drive the equivalent of the era of personal computing this time.
Space.
You don't want to wait 3-22 minutes for a ping from Mars.
I'm not sure if the handful of people in space stations are a big enough market to drive such changes.
Centralized only became mainstream when everything started to be offered "for free". When it was buy or pay recurrently more often the choice was to buy.
There are no longer options to buy. Everything is a subscription
Between mobilephone service including SMS and an ISP service which usually include mail I don't see the need for any hosted service.
There are FOSS alternatives for about everything for hobbyist and consumer use.
There are no FOSS alternatives for consumer use unless the consumer is an IT pro or a developer. Regular people can’t use most open source software without help. Some of it, like Linux desktop stuff, has a nice enough UI that they can use it casually but they can’t install or configure or fix it.
Making software that is polished and reliable and automatic enough that non computer people can use it is a lot harder than just making software. I’d say it’s usually many times harder.
I don't think that is a software issue but a social issue nowadays. FOSS alternatives have become quite OK in my opinion.
If computers came with Debian, Firefox and Libre Office preinstalled instead of only W11, Edge and with some Office 365 trail, the relative difficulty would be gone I think.
Same thing with most IT departments only dealing with Windows in professional settings. If you even are allowed to use something different you are on your own.
I think people have seen enough of this 'free' business model to know the things being sold for free are in fact, not.
Some people, but a majority see it as free. Go to your local town center and randomly poll people how much they pay for email or google search, 99% will say it is free and stop there.
Privacy, reliable access when not connected to the web, the principal of decentralizing for some. Less supply chain risk for private enterprise.
> We now have high speed internet access everywhere
This is such a HN comment illustrating how little your average HN knows of the world beyond their tech bubble. Internet everywhere, you might have something of a point. But "high speed internet access everywhere" sounds like "I haven't travelled much in my life".
I don't know, I think you're conflating content streaming with central compute.
Also, is percentage of screentime the relevant metric? We moved TV consumption to the PC, does that take away from PCs?
Many apps moved to the web but that's basically just streamed code to be run in a local VM. Is that a dumb terminal? It's not exactly local compute independent...
Nah, your parent comment has a valid point.
Nearly entirety of the use cases of computers today don't involve running things on a 'personal computer' in any way.
In fact these days, every one kind of agrees as little as hosting a spreadsheet on your computer is a bad idea. Cloud, where everything is backed up is the way to go.
But again, that's conflating web connected or even web required with mainframe compute and it's just not the same.
PC was never 'no web'. No one actually 'counted every screw in their garage' as the PC killer app. It was always the web.
You know that the personal computer predates the web by quite a few years?
Sure, I was too hyperbolic. I simply meant connecting to the web didn't make it not a PC.
The web really pushed adoption, much more than a person computation machine. It was the main use case for most folks.
This. Although briefly, there was at least a couple of years of using pc's without an internet connection. It's unthinkable now. And even back then, when you blinked with your eyes this time period was over.
That was a pretty long blink?
The personal computer arguably begins with VisiCalc in 1979.
> Through the 1970s, personal computers had proven popular with electronics enthusiasts and hobbyists, however it was unclear why the general public might want to own one. This perception changed in 1979 with the release of VisiCalc from VisiCorp (originally Personal Software), which was the first spreadsheet application.
https://en.wikipedia.org/wiki/History_of_personal_computers#...
Mainstream use of the web really took off in the second half of the 1990s. Arbitrarily, let's say with the release of Windows 95. That's a quarter of a century you'd be blinking for.
One of the actual killer apps was gaming. Which still "happens" mostly on the client, today, even for networked games.
Yet the most popular games are online-only and even more have their installation base's copies of the game managed by an online-first DRM.
That's true, but beside the point: even online only games or those gated by online DRM are not streamed or resemble a thin client architecture.
That exists, too, with GeForce Now etc, which is why I said mostly.
This is just factually inaccurate.
Please provide a more comprehensive response. I suppose I should be more specific as well.
Some of the online only games I am thinking of are CoD, Fortnite, LoL and Minecraft. The online-first DRM I am thinking of is Steam.
Even half of your list is not "online only."
In time Mainframes of this age will make a come back.
This whole idea that you can connect lots of cheap low capacity boxes and drive down compute costs is already going away.
In time people will go back to thinking compute as a variable of time taken to finish processing. That's the paradigm in the cloud compute world- you are billed for the TIME you use the box. Eventually people will just want to use something bigger that gets things done faster, hence you don't have to rent them for long.
It's also interesting that computing capacity is no longer discussed as instructions per second, but as Giga Watts.
Umm... I had a PC a decade before the web was invented, and I didn't even use the web for like another 5 years after it went public ("it's an interesting bit of tech but it will obviously never replace gopher...")
The killer apps in the 80s were spreadsheets and desktop publishing.
> I don't know, I think you're conflating content streaming with central compute.
Would you classify eg gmail as 'content streaming'?
But gmail is also a relatively complicated app, much of which runs locally on the client device.
It is true that browsers do much more computation than "dumb" terminals, but there are still non-trivial parallels. Terminals do contain a processor and memory in order to handle settings menus, handle keyboard input and convert incoming sequences into a character array that is then displayed on the screen. A terminal is mostly useless without something attached to the other side, but not _completely_ useless. You can browse the menus, enable local echo, and use device as something like a scratchpad. I once drew up a schematic as ascii art this way. The contents are ephemeral and you have to take a photo of the screen or something in order to retain the data.
Web browsers aren't quite that useless with no internet connection, some sites do offer offline capabilities (for example gmail). but even then, the vast majority of offline experiences exist to tide the user over until network can be re-established, instead of truly offering something useful to do locally. Probably the only mainstream counter-examples would be games.
It's still a SAAS, with components that couldn't be replicated client-side, such as AI.
Google's own Gemma models are runnable locally on a Pixel 9 Max so some lev of AI is replicatable client side. As far as Gmail running locally, it wouldn't be impossible for Gmail to be locally hosted and hit a local cache which syncs with a server only periodically over IMAP/JMAP/whatever if Google actually wanted to do it.
Yes, but seems like a lot of hassle for not much gain (for Google).
The gain, as far as local AI goes for Google, is that, at Google scale, the CPU/GPU time to run even a small model like Gemma will add up across Gmail's millions of users. If clients have the hardware for it (which Pixel 9's have) it means Gmail's servers aren't burning CPU/GPU time on it.
As far as how Gmail's existing offline mode works, I don't know.
To make the discussion easier, we can look at the state of GMail before widespread AI was a thing.
(I have to use some weaselwording here, because GMail had decent spam detection since basically forever, and whether you call that AI or not depends on where we have shifted the goalposts at the moment.)
Heh that brings to mind that old chestnut, something like "We call it AI until it works, then we call it machine learning." Google's running the servers tho, so the assumption is the servers would be running the spam filter against it before the user. In my mind, anyway. So it's fair to point out that Google also has the compute resources to not have to care. To the hypothetical though, could Google make Gmail a local-first? I say yes. Will they? I doubt it.
Right. But does it matter whether computation happens on the client or server? Probabaly on both in the end.
But yes I am looking forward to having my own LMS on my PC which only I have access to.
Well, app code is streamed, content is streamed. The app code is run locally. Content is pulled periodically.
The mail server is the mail server even for Outlook.
Outlook gives you a way to look through email offline. Gmail apps and even Gmail in Chrome have an offline mode that let you look through email.
It's not easy to call it fully offline, nor a dumb terminal.
Oh, GMail is definitely a cloud offering---even if they have some offline functionality.
I was just probing the 'content _streaming_' term. As you demonstrate, you'd have to squint really hard to describe GMail as content streaming.
'Offline' vs 'content streaming' is a false dichotomy. There's more different types of products and services.
(Which reminds me a bit of crypto-folks calling everything software that's not in crypto "web2", as if working on stodgy backends in a bank or making Nintendo Switch games has anything to do with the web at all.)
Ok sure but there's plenty of simple video streaming in total screen time, which was the context I was replying to. I never claimed it was a dichotomy, simply a large part of screen time that clearly skews the analysis.
Yes, it's a large part of screen time at the moment.
> using the computer as a dumb terminal to access centralized services "in the cloud"
Our personal devices are far from thin clients.
Depends on the app, and the personal device. Mobile devices are increasingly thin clients. Of course hardware-wise they are fully capable personal computers, but ridiculous software-imposed limitations make that increasingly difficult.
"Thin" can be interpreted as relative, no?
I think it depends on if you see the browser for content or as a runtime environment.
Maybe it depends on the application architecture...? I.e., a compute-heavy WASM SPA at one end vs a server-rendered website.
Or is it an objective measure?
But that is what they are mostly used for.
On phones, most of the compute is used to render media files and games, and make pretty animated UIs.
The text content of a weather app is trivial compared to the UI.
Same with many web pages.
Desktop apps use local compute, but that's more a limitation of latency and network bandwidth than any fundamental need to keep things local.
Security and privacy also matter to some people. But not to most.
Speak for yourself. Many people don't daily-drive anything more advanced than an iPad.
IPads are incredibly advanced. Though I guess you mean they don't use anything that requires more sophistication from the user (or something like that)?
The Ipad is not a thin client, is it?
It is, for the vast majority of users.
Turn off internet on they iPad and see how many apps that people use still work.
I'm not questioning whether the Ipad can be used as a client in some capacity, or whether people tend to use it as a client. I question whether the Ipad is a thin client. The answer to that question doesn't lie in how many applications require an internet connection, but in how many applications require local computational resources.
The Ipad is a high performance computer, not just because Apple think that's fun, but out of necessity given its ambition: the applications people use on it require local storage and rather heavy local computation. The web browser standards if nothing else have pretty much guaranteed that the age of thin clients is over: a client needs to supply a significant amount of computational resources and storage to use the web generally. Not even Chromebooks will practically be anything less than rich clients.
Going back to the original topic (and source of the analogy), IOS hosts an on-device large language model.
As with everything, the lines are a bit blurred these days. We may need a new term for these devices. But despite all the compute and storage and on-device models these supercomputers are barely a step above thin clients.
No, its a poor anology, I'm old enough to have used a Wyse terminal. That's what I think of when I hear dumb terminal. It was dumb.
Maybe a PC without a hard drive (PXE the OS), but if it has storage and can install software, its not dumb.
We may want a new term for our devices :) https://news.ycombinator.com/item?id=45808654
I mean, Chromebooks really aren't very far at all from thin clients. But even my monster ROG laptop when it's not gaming is mostly displaying the results of computation that happened elsewhere
There are more PCs and serious home computing setups today than there were back then. There are just way way way more casual computer users.
The people who only use phones and tablets or only use laptops as dumb terminals are not the people who were buying PCs in the 1980s and 1990s, or they were they were not serious users. They were mostly non-computer-users.
Non-computer-users have become casual consumer level computer users because the tech went mainstream, but there's still a massive serious computer user market. I know many people with home labs or even small cloud installations in their basements, but there are about as many of them as serious PC users with top-end PC setups in the late 1980s.
I dislike the view of individuals as passive sufferers of the preferences of big corporations.
You can and people do self-host stuff that big tech wants pushed into the cloud.
You can have a NAS, a private media player, Home Assistant has been making waves in the home automation sphere. Turns out people don't like buying overpriced devices only to have to pay a $20 subscription, and find out their devices don't talk to each other, upload footage inside of their homes to the cloud, and then get bricked once the company selling them goes under and turns of the servers.
This. And the hordes of people reacting with some explanation for why this is. The 'why' is not the point, we already know the 'why'. The point is that you can if you want. Might not be easy, might not be convenient, but that's not the point. No one has to ask someone else for permission to use other tech than big tech.
The explanation of 'why' is not an argument. Big tech is not making it easy != it's impossible. Passive sufferers indeed.
Edit: got a website with an RSS feed somewhere maybe? I would like to follow more people with a point of view like yours.
You can dislike it but it doesn't make it less true and getting truer.
You can likewise host models if you so choose. Still the vast majority of people use online services both for personal computing or for LLMs.
Things are moving this way because it’s convenient and easy and most people today are time poor.
I think it has more to do with the 'common wisdom' dictating that this is the way to do it, as 'we've always done it like this'.
Which might even be true, since cloud based software might offer conveniences that local substitutes don't.
However this is not an inherent property of cloud software, its just some effort needs to go into a local alternative.
That's why I mentioned Home Assistant - a couple years ago, smart home stuff was all the rage, and not only was it expensive, the backend ran in the cloud, and you usually paid a subscription for it.
Nowadays, you can buy a local Home Assistant hub (or make one using a Pi) and have all your stuff only connect to a local server.
The same is true for routers, NAS, media sharing and streaming to TV etc. You do need to get technical a bit, but you don't need to do anything you couldn't figure out by following a 20 minute Youtube video.
I look forward to a possibility where the dumb terminal is less centralized in the cloud, and more how it seems to work in the expanse. They all have hand terminals that seem to automatically interact with the systems and networks of the ship/station/building they're in. Linking up with local resources, and likely having default permissions set to restrict weird behavior.
Not sure it could really work like that IRL, but I haven't put a ton of thought into it. It'd make our always-online devices make a little more sense.
But for a broader definition of "personal computer", the number of computers we have has only continued to skyrocket - phones, watches, cars, TVs, smart speakers, toaster ovens, kids' toys...
I'm with GP - I imagine a future when capable AI models become small and cheap enough to run locally in all kinds of contexts.
https://notes.npilk.com/ten-thousand-agents
Depending on how you are defining AI models, they already do. Think of the $15 security camera that can detect people and objects. That is AI model driven. LLM's are another story, but smaller, less effective ones can and do already run at the edge.
I think that speaks more to the fact that software ate the world, than locality of compute. It's a breadth first, depth last game.
Makes me want to unplug and go back to offline social media. That's a joke. The dominant effect was networked applications getting developed, enabling community, not a shift back to client terminals.
Once up on a time social media was called Usenet and worked offline in a dedicated client with a standard protocol. You only went online to download and send messages, but could then go offline and read them in an app of your choice.
Web2.0 discarded the protocol approach and turned your computer into a thin client that does little more than render webapps that require you to be permanently online.
> Once up on a time social media was called Usenet and worked offline in a dedicated client with a standard protocol.
There was also FidoNet with offline message readers.
> called Usenet and worked offline
People must have been pretty smart back then. They had to know to hang up the phone to check for new messages.
I guess we're in the kim-1 era of local models, or is that already done?
That 'average' is doing a lot of work to obfuscate the landscape. Open source continues to grow (indicating a robust ecosystem of individuals who use their computers for local work) and more importantly, the 'average' looks like it does not necessarily due to a reduction in local use, but to an explosion of users that did not previously exist (mobile first, SAAS customers, etc.)
The thing we do need to be careful about is regulatory capture. We could very well end up with nothing but monolithic centralized systems simply because it's made illegal to distribute, use, and share open models. They hinted quite strongly that they wanted to do this with deepseek.
There may even be a case to be made that at some point in the future, small local models will outperform monoliths - if distributed training becomes cheap enough, or if we find an alternative to backprop that allows models to learn as they infer (like a more developed forward-forward or something like it), we may see models that do better simply because they aren't a large centralized organism behind a walled garden. I'll grant that this is a fairly polyanna take and represents the best possible outcome but it's not outlandishly fantastic - and there is good reason to believe that any system based on a robust decentralized architecture would be more resilient to problems like platform enshittification and overdeveloped censorship.
At the end of the day, it's not important what the 'average' user is doing, so long as there are enough non-average users pushing the ball forward on the important stuff.
We already have monolithic centralised systems.
Most open source development happens on GitHub.
You'd think non-average developers would have noticed their code is now hosted by Microsoft, not the FSF. But perhaps not.
The AI end game is likely some kind of post-Cambrian, post-capitalist soup of evolving distributed compute.
But at the moment there's no conceivable way for local and/or distributed systems to have better performance and more intelligence.
Local computing has latency, bandwidth, and speed/memory limits, and general distributed computing isn't even a thing.
I can't imagine a universe where a small mind with limited computing resources has an advantage against a datacenter mind, no matter the architecture.
The small mind could have an advantage if it is closer or more trustworthy to users.
It only has to be good enough to do what we want. In the extreme, maybe inference becomes cheap enough that we ask “why do I have to wake up the laptop’s antenna?”
I would like to have a personal AI agent which basically has a copy of my knowledge, a reflection of me, so it could help me mupltiply my mind.
I don't want to send sensitive information to a data center, I don't want it to leave my machine/network/what have you. Local models can help in that department.
You could say the same about all self-hosted software, teams with billions of dollars to produce and host SaaS will always have an advantage over smaller, local operations.
Abundant resources could enable bad designs. I could in particular see a lot of commercial drive for huge models that can solve a bazillion different use cases, but aren't efficient for any of them.
There might be also local/global bias strategies. A tiny local model trained on your specific code/document base may be better aligned to match your specific needs than a galaxy scale model. If it only knows about one "User" class, the one in your codebase, it might be less prone to borrowing irrelevant ideas from fifty other systems.
The advantage it might have won't be in the form of "more power", it would be in the form of "not burdened by sponsored content / training or censorship of any kind, and focused on the use-cases most relevant to the individual end user."
We're already very, very close to "smart enough for most stuff". We just need that to also be "tuned for our specific wants and needs".
The only difference is latency.
Universes like ours where the datacentre mind is completely untrustworthy.
Even the most popular games (with few exceptions) present as relatively dumb terminals that need constant connectivity to sync every activity to a mainframe - not necessarily because it's an MMO or multiplayer game, but because it's the industry standard way to ensure fairness. And by fairness, of course, I mean the optimization of enforcing "grindiness" as a mechanism to sell lootboxes and premium subscriptions.
And AI just further normalizes the need for connectivity; cloud models are likely to improve faster than local models, for both technical and business reasons. They've got the premium-subscriptions model down. I shudder to think what happens when OpenAI begins hiring/subsuming-the-knowledge-of "revenue optimization analysts" from the AAA gaming world as a way to boost revenue.
But hey, at least you still need humans, at some level, if your paperclip optimizer is told to find ways to get humans to spend money on "a sense of pride and accomplishment." [0]
We do not live in a utopia.
[0] https://www.guinnessworldrecords.com/world-records/503152-mo... - https://www.reddit.com/r/StarWarsBattlefront/comments/7cff0b...
I imagine there are plenty of indie single-player games that work just fine offline. You lose cloud saves and achievements, but everything else still works.
Funny you would pick this analogy. I feel like we’re back in the mainframe era. A lot of software can’t operate without an internet connection. Even if in practice they execute some of the code on your device, a lot of the data and the heavyweight processing is already happening on the server. Even basic services designed from the ground up to be distributed and local first - like email (“downloading”) - are used in this fashion - like gmail. Maps apps added offline support years after they launched and still cripple the search. Even git has GitHub sitting in the middle and most people don’t or can’t use git any other way. SaaS, Electron, …etc. have brought us back to the mainframe era.
It's always struck me as living in some sort of bizaro world. We now have these super powerful personal computers, both handheld (phones) and laptops (My M4 Pro smokes even some desktop class processors) and yet I use all this powerful compute hardware to...be a dumb terminal to someone else's computer.
I had always hoped we'd do more locally on-device (and with native apps, not running 100 instances of chromium for various electron apps). But, it's hard to extract rent that way I suppose.
What's truly wild when you think about it, is that the computer on the other end is often less powerful than your personal laptop.
I access websites on a 64gb, 16 core device. I deploy them to a 16gb, 4 core server.
Yes, but your computer relies on dozens (hundreds?) of servers at any given time.
I don't even understand why computer and phone manufacturers even try to make their devices faster anymore, since for most computing tasks, the bottleneck is all the data that needs to be transferred to and from the modern version of the mainframe.
There are often activities that do require compute though. My last phone upgrade was so Pokemon Go would work again, my friend upgrades for the latest 4k video or similar.
Consumers care about battery life.
Yet manufacturers give us thinner and thinner phones every year (instead of using that space for the battery), and make it difficult to swap out batteries which have degraded.
> make it difficult to swap out batteries which have degraded.
That's the part that pisses me off the most. They all claim it's for the IP68, but that's bullshit. There's plenty of devices with removable backs & batteries that are IP68.
My BlackBerry bold 9xxx was 10mm thin. the iPhone 17 Pro Max is 8.75. You aren't going to notice the 1.3mm of difference, and my BlackBerry had a user replaceable battery, no tools required just pop off the back cover.
The BlackBerry was also about 100 grams lighter.
The non-user removable batteries and unibody designs are purely for planned obsolescence, nothing else.
Also when a remote service struggle I can switch to do something else. When a local software struggles it brings my whole device to its knees and I can't do anything.
And providers count their capacity in Giga-watts.
I think people have been finding more compelling use cases for the fact that information systems can be multi-player now than for marginal FLOPS. Client-server is just a very effective way of organizing multi-player information systems.
yet I use all this powerful compute hardware to...animate liquid glass
> A lot of software can’t operate without an internet connection
Or even physical things like mattresses, according to discussions around the recent AWS issues.
Why would companies sell you the golden goose when they can instead sell you an egg every day?
> Why would companies sell you the golden goose when they can instead sell you an egg every day?
Because someone else can sell the goose and take your market.
Apple is best aligned to be the disruptor. But I wouldn’t underestimate the Chinese government dumping top-tier open-source models on the internet to take our tech companies down a notch or ten.
By that logic none of us should be paying monthly subscriptions for anything because obviously someone would disrupt that pricing model and take business away from all the tech companies who are charging it? Especially since personal computers and mobile devices get more and more powerful and capable with every passing year. Yet subscriptions also get more prevalent every year.
If Apple does finally come up with a fully on-device AI model that is actually useful, what makes you think they won't gate it behind a $20/mo subscription like they do for everything else?
> By that logic none of us should be paying monthly subscriptions for anything because obviously someone would disrupt that pricing model and take business away from all the tech companies who are charging it?
Non sequitur.
If a market is being ripped off by subscription, there is opportunity in selling the asset. Vice versa: if the asset sellers are ripping off the market, there is opportunity to turn it into a subscription. Business models tend to oscillate between these two for a variety of reasons. Nothing there suggets one mode is infinitely yielding.
> If Apple does finally come up with a fully on-device AI model that is actually useful, what makes you think they won't gate it behind a $20/mo subscription like they do for everything else?
If they can, someone else can, too. They can make plenty of money selling it straight.
> If a market is being ripped off by subscription, there is opportunity in selling the asset.
Only in theory. Nothing beats getting paid forever.
> Business models tend to oscillate between these two for a variety of reasons
They do? AFAICT everything devolves into subscriptions/rent since it maximizes profit. It's the only logical outcome.
> If they can, someone else can, too.
And that's why companies love those monopolies. So, no... other's can't straight up compete against a monopoly.
Because they need to displace open AI users, or open AI will steer their trajectory towards Apple at some point.
What on-device app does Apple charge a subscription for?
Sure, the company that launched iTunes and killed physical media, then released a phone where you can't install apps ("the web is the apps") will be the disruptor to bring back local computing to users...
> Apple is best aligned to be the disruptor.
It's this disruptor Apple in the room with us now?
Apple's second biggest money source is services. You know, subscriptions. And that source keeps growing: https://sixcolors.com/post/2025/10/charts-apple-caps-off-bes...
It's also that same Apple that fights tooth and nail every single attempt to let people have the goose or even the promise of a goose. E.g. by saying that it's entitled to a cut even if a transaction didn't happen through Apple.
Unfortunately, most people just want eggs, not the burden of actually owning the goose.
Putting a few boots in Taiwan would also make for a profitable short. Profitable to the tune of several trillion dollars. Xi must be getting tempted.
It's a lot more complicated than that. They need to be able to take the island very quickly with a decapitation strike, while also keeping TSMC from being sabotaged or destroyed, then they need to be able to weather a long western economic embargo until they can "break the siege" with demand for what they control along with minor good faith concessions.
It's very risky play, and if it doesn't work it leaves China in a much worse place than before, so ideally you don't make the play unless you're already facing some big downside, sort of as a "hail Mary" move. At this point I'm sure they're assuming Trump is glad handing them while preparing for military action, they might even view invasion of Taiwan as defensive if they think military action could be imminent anyhow.
Destroying TSMC or knowing it would be sabotaged would pretty much be the point of the operation. Would take 48 hours and they could be out of there again and say “ooops sorry” at the UN.
Hard disagree. They need chips bad, and it's the US defense position that TSMC be destroyed if possible in the event of successful Chinese invasion. They also care about reunification on principle, and an attack like that without letting them force "One China" on the Taiwanese in the aftermath would just move them farther from that goal.
> then they need to be able to weather a long western economic embargo until they can "break the siege" with demand for what they control along with minor good faith concessions
And you know we'd be potting their transport ships, et cetera, from a distance the whole time, all to terrific fanfare. The Taiwan Strait would become the new training ground for naval drones, with the targets being almost exclusively Chinese.
I worked with the Taiwanese Military, that's their dream scenario but the reality is they're scared shitless that the Chinese will decapitate them with massive air superiority. Drones don't mean shit without C2.
> they're scared shitless that the Chinese will decapitate them with massive air superiority
Taiwan fields strong air defenses backed up by American long-range fortifications.
The threat is covert decapitation. A series of terrorist attacks carried out to sow confusion while the attack launches.
Nevertheless, unless China pulls off a Kabul, they’d still be subject to constant cross-Strait harassment.
China has between 5:1 and 10:1 advantage depending on asset class. If not already on standby, US interdiction is ~48 hours. For sure China is going to blast on all fronts, so cyber and grid interruptions combined with shock and awe are definitely gonna be a thing. It's not a great setup for Taiwan.
You could say the same thing about Computers when they were mostly mainframe. I am sure someone will figure out how to make it commoditized just like personal computers and internet.
An interesting remark: in the 1950s-1970s, mainframes were typically rented rather than sold.
It looks to me like the personal computer area is over. Everything is in the cloud and accessed through terminals like phones and tablets.
And notably, those phones and tablets are intentionally hobbled by the device owners (Apple, Google) who do everything they can to ensure they can't be treated like personal computing devices. Short of regulatory intervention, I don't see this trend changing anytime soon. We're going full on in the direction of more locked down now that Google is tightening the screws on Android.
Because someone else will sell it to you if they dont.
Because companies are not some monolith, all doing identical things forever. If someone sees a new angle to make money, they'll start doing it.
Data General and Unisys did not create PCs - small disrupters did that. These startups were happy to sell eggs.
They didn't create them, but PC startups like Apple and Commodore only made inroads into the home -- a relatively narrow market compared to business. It took IBM to legitimize PCs as business tools.
Well if there's at least one competitor selling golden geese to consumers the rest have to adapt.
Assuming consumers even bother to set up a coop in their living room...
Your margin is my opportunity. The more expensive centralized models get the easier it is for distributed models to compete.
Exactly! It's a rent-seeking model.
> I look forward to the "personal computing" period, with small models distributed everywhere...
Like the web, which worked out great?
Our Internet is largely centralized platforms. Built on technology controlled by trillion dollar titans.
Google somehow got the lion share of browser usage and is now dictating the direction of web tech, including the removal of adblock. The URL bar defaults to Google search, where the top results are paid ads.
Your typical everyday person uses their default, locked down iPhone or Android to consume Google or Apple platform products. They then communicate with their friends over Meta platforms, Reddit, or Discord.
The decentralized web could never outrun money. It's difficult to out-engineer hundreds of thousands of the most talented, most highly paid engineers that are working to create these silos.
Ok, so Brave Browser exists - if you download, you will see 0 ads on the internet, I've never really seen ads on the internet - even in the before brave times.
Fr tho, no ads - I'm not making money off them, I've got no invite code for you, I'm a human - I just don't get it. I've probably told 500 people about Brave, I don't know any that ever tried it.
I don't ever know what to say. You're not wrong, as long as you never try to do something else.
Brave is just a rebranded Chrome. By using it you’re still endorsing Google’s control of the web.
I was gonna say this. If Google decides to stop developing chromium then Brave is left with very few choices.
As someone who has been using brace since it was first announced and very tightly coupled to the BAT crypto token I must say it is much less effective nowadays.
I often still see a load of ads and also regularly have to turn off the shields for some sites.
How?? I watch Hulu ad supported without ads, YouTube no ads ever, screen never goes black, Viki ad supported free package - never seen an ad, Paramount ad package - never seen an ad.
I never have to turn off shields - on one hand the number of times I've had to do that.
Maybe I have something additional installed I don't know.
If everyone used Brave, Google wouldn't be a multi-trillion dollar company pulling revenues that dwarf many countries.
Or rather, they'd block Brave.
I agree man, it's depressing.
When the consumer decides to discover my site and fund federated and P2P infrastructure, they can have a seat at the table.
Selling fertile geese was a winning and proven business biz model for a very long time.
Selling eggs is better how?
We have a ton of good, small models. The issues are:
1. Most people don't have machines that can run even midsized local models well
2. The local models are nearly as good as the frontier models for a lot of use cases
3. There are technical hurdles to running local models that will block 99% of people. Even if the steps are: download LM Studio and download a model
Maybe local models will get so good that they cover 99% of normal user use cases and it'll be like using your phone/computer to edit a photo. But you'll still need something to make it automatic enough that regular people use it by default.
That said, anyone reading this is almost certainly technical enough to run a local model. I would highly recommend trying some. Very neat to know it's entirely run from your machine and seeing what it can do. LM Studio is the most brainless way to dip your toes in.
As the hype is dying down it's becoming a little bit clearer that AI isn't like blockchain and might be actually useful (for non generative purposes at least)
I'm curious what counts as a midsize model; 4B, 8B, or something larger/smaller?
What models would you recommend? I have 12GB of vram so anything larger than 8B might be really slow, but i am not sure
My take:
Large: Requires >128GB VRAM
Medium: 32-128GB VRAM
Small: 16GB VRAM
Micro: Runs on a microcontroller or GPUs with just 4GB of VRAM
There's really nothing worthwhile for general use cases that runs in under 16GB (from my testing) except a grammar-checking model that I can't remember the name of at the moment.
gpt-oss:20b runs on 16GB of VRAM and it's actually quite good (for coding, at least)! Especially with Python.
Prediction: The day that your average gaming PC comes with 128GB of VRAM is the day developers will stop bothering with cloud-based AI services. gpt-oss:120b is nearly as good as gpt5 and we're still at the beginning of the AI revolution.
It can depend on your use case. Are you editing a large code base and will thus make lots of completion requests with large contexts?
Try WebLLM - it's pretty decent and all in-browser/offline even for light tasks, 1B-1.5B models like Qwen2.5-Coder-1.5B-Instruct. I put together a quick prototype - CodexLocal.com but you can essentially a local nginx and use webllm as an offline app. Of course, you can just use Ollama / LM Studio but that would require a more technical solution
I like to think of it more broadly, and that we are currently in the era of the first automobile. [0]
LLMs are the internal combustion engine, and chatbot UIs are at the "horseless carriage" phase.
My personal theory is that even if models stopped making major advancements, we would find cheaper and more useful ways to use them. In the end, our current implementations will look like the automobile pictured below.
[0] https://group.mercedes-benz.com/company/tradition/company-hi...
I'm not a big google fan, but I really like the "Google AI Edge Gallery" android app [0]. In particular, I've been chatting with the "Gemma-3n-E2B-it" model when I don't have an internet connection, and it's really decent!
[0] https://play.google.com/store/apps/details?id=com.google.ai....
Don't we already have small models highly distributed?
We do, but the vast majority of users interact with centralised models from Open AI, Google Gemini, Grok...
I'm not sure we can look forward to self-hosted models ever being mainstream.
Like 50% of internet users are already interacting with one of these daily.
You usually only change your habit when something is substantially better.
I don't know how free versions are going to be smaller, run on commodity hardware, take up trivial space and ram etc, AND be substantially better
> I'm not sure we can look forward to self-hosted models ever being mainstream.
If you are using an Apple product chances are you are already using self-hosted models for things like writing tools and don't even know it.
The "enshittification" hasn't happened yet. They'll add ads and other gross stuff to the free or cheap tiers. Some will continue to use it, but there will be an opportunity for self-hosted models to emerge.
> Like 50% of internet users are already interacting with one of these daily. You usually only change your habit when something is substantially better.
No, you usually only change your habit when the tools you are already using are changed without consulting you, and the statistics are then used to lie.
You make a fair point, I'm just hoping this will happen, but not confident either to be frank.
Because small models are just not that good.
The vast majority won't switch until there's a 10x use case. We know they are coming. Why bother hopping?
I use gpt4all and have downloaded some models. It's not that hard, and doesn't take a huge system. It's pretty easy if you use models on their list, it's a bit more work to find the right chat interactive script (forgot what they are called) for your class of model if you downloaded one that's not on the list, but not that hard. I have 8GB VRAM, and it's slow but they work. I like that it's local and private, but then I don't use them for much other than as an oddity that is fun to interact with.
Dial-up + mainframe. Mainframe from POV as silos, dial-up internet as the speed we have now when looking back to 2025 in 2035.
Mainframes still exist, and they actually make a lot of sense from physics perspective. It's good idea to run transactions in a big machine rather than distributed, the latter is less energy efficient.
I think the misconception is that things cannot be overpriced for reasons other than inefficiency.
We are also in the mainframe period of computing, with large centralised cloud services.
I actually think we are much closer to the sneaker era of shoes, or the monorail era of public transit.
this -- chips are getting fast enough both arm n x86. unified memory architecture means we can get more ram on devices at faster throughput. we're already seeing local models - just that their capability is limited by ram.
I think we are in the dotcom boom era where investment is circular and the cash investments all depend on the idea that growth is infinite.
Just a bunch of billionaires jockeying for not being poor.
ollama and other peojects already make this possible
> "personal computing" period
The period when you couldn't use Linux as your main OS because your organization asked for .doc files?
No thanks.
I actually don’t look forward to this period. I have always been for open source software and distributism — until AI.
Because if there’s one thing worse than governments having nuclear weapons, it’s everyone having them.
It would be chaos. And with physical drones and robots coming, it woukd be even worse. Think “shitcoins and memecoins” but unlike those, you don’t just lose the money you put in and you can’t opt out. They’d affect everyone, and you can never escape the chaos ever again. They’d be posting around the whole Internet (including here, YouTube deepfakes, extortion, annoyance, constantly trying to rewrite history, get published, reputational destruction at scale etc etc), and constant armies of bots fighting. A dark forest.
And if AI can pay for its own propagation via decentralized hosting and inference, then the chance of a runaway advanced persistent threat compounds. It just takes a few bad apples, or even practical jokers, to unleash crazy stuff. And it will never be shut down, just build and build like some kind of kessler syndrome. And I’m talking about with just CURRENT AI agent and drone technology.
I mean, people can self-host plenty off of a 5090, heck even Macs with enough RAM can run larger models that I can't run on a 5090.
Imagine small models on a cheap chip that can be added to anything (alarm clock, electric toothbrush, car keys...)
Funny how this guy thinks he knows exactly what's up with AI, and how "others" are "partly right and wrong." Takes a bit of hubris to be so confident. I certainly don't have the hubris to think I know exactly how it's all going to go down.
How about a vague prediction that covers all scenarios? XD
*ahem* It's gonna be like every other tool/societal paradigm shift like the smartphone before this, and planes/trains/cars/ships/factories/electricity/oil/steam/iron/bronze etc. before that:
• It'll coalesce into the hands of a few corporations.
• Idiots in governments won't know what the fuck to do with it.
• Lazy/loud civvies will get lazier/louder through it.
• There'll be some pockets of individual creativity and freedom, like open source projects, that will take varying amounts of time to catch on in popularity or fade away to obscurity.
• One or two killer apps that seem obvious but nobody thought of, will come out of nowhere from some nobody.
• Some groups will be quietly working away using it to enable the next shift, whether they know it or not.
• Aliens will land turning everything upside down. (I didn't say when)
Forgot:
• Militaries will want to kill everyone with it.
But do you have the audacity to be wrong?
Yeah that's interesting, good perspective
The problem is that the bubble people are so unimaginative, similar to Krugman, that those who have any inkling of an imagination can literally feel like visionaries compared to them. I know I’m describing Dunning-Krueger, but so be it, the bubble people are very very wrong. It’s like, man, they really are unable to imagine a very real future.
It’s a weird comparison since internet in the dial-up age was a bubble, are you saying the hype machine for AI is in fact smaller than the internet? Are you implying that AI will in fact grow that much more slowly and sustainably than the internet, despite trillions of investment?
Do you think Sam Altman, Jeff Bezos, and Mark Zuckerberg are all wrong saying that we’re in a bubble? Do they “lack imagination?”
Also? What do I need imagination for, isn’t that what AI does now?
That’s a sharp and layered question, and I think you’re cutting right to the heart of the current tension around AI.
Not only was it a bubble but we actually had the internet. We didn't have to use our imaginations and build for the technology yet to be invented and we still had a massive damaging bubble.
Even beyond that, the Soviet Union had just collapsed. The US had a balanced budget and 10X less debt. Globalization was just getting started. China was nothing compared to what it was today. It was the absolute most stable time of my life. Basically the opposite of 2025.
"This time things are different because AI blah blah blah".
When you boil it down, all bubbles are based around the idea of using your imagination to image how "this time it is different".
The people who don't think this is a bubble are not seeing that their imagination is a bug and not a feature here.
I find the argument for the bubble to be extremely straightforward.
Currently, investment into AI exceeds the dot-com bubble by a factor of 17. Even in the dot-com era, the early internet was already changing media and commerce in fundamental ways. November is the three-year anniversary of ChatGPT. How much economic value are they actually creating? How many people are purchasing AI-generated goods? How much are people paying for AI-provided services? The value created here would have to exceed what the internet was generating in 2000 by a factor of 17 (which seems excessive to me) to even reach parity with the dot-com bubble.
"But think where it'll be in 5 years"—sure, and let's extrapolate that based on where it is now compared to where it was 3 years ago. New models present diminishing returns. 3.5 was groudbreaking; 4 was a big step forward; 5 is incremental. I won't deny that LLMs are useful, and they are certainly much more productized now than they were 3 years ago. But the magnitude of our collective investment in AI requires that a huge watershed moment be just around the corner, and that makes no sense. The watershed moment was 3 years ago. The first LLMs created a huge amount of potential. Now we're realizing those gains, and we're seeing some real value, but things are also tapering off.
Surely we will have another big breakthrough some day—a further era of AI which brings us closer to something like AGI—but there's just no reason to assume AGI will crop up in 2027, and nothing less that that can produce the ROI that such enormous valuations will eventually, inexorably, demand.
I don’t get why people find it so hard to understand that a technology can be value-additive and still be in a position of massive overinvestment. Every generation of Californians seeks to relive the 1848 gold rush, spending millions excavating rivulets for mere ounces of (very real!) gold.
Exactly this. The future impact of AI and the financial credibility of OpenAI as a business are completely distinct.
Not to mention the 1848 gold rush pretty destroyed the existing society, culture and businesses:
https://en.wikipedia.org/wiki/California_gold_rus
Not to mention thousands of native inhabitants getting killed or enslaved:
https://en.wikipedia.org/wiki/California_genocide
What is AGI in your mind? Let's take someone who once upon a time was responsible for grading papers. As far as that person is concerned, AGI has arrived for their profession (it arrived nearly two years ago for them). You'll never be better than something that has read every book ever and can write better than you. AGI will come in tranches. Are you really going to hire that developer because you need extra manpower to stand up test coverage? No, so as far as that developer is concerned, AGI has arrived for that part of their professional career.
The bet is not that there will be this one seminal moment of AGI where all the investment will make sense. The bet is that it has already showed up if you look for specific things and will continue to do so. I wouldn't bet against the idea that LLMs will introduce itself to all jobs, one at a time. Reddit moderators, for example, will meet AGI (as far as they know, their entire world being moderating) sooner than say, I don't know, a Radiologist.
The universe of people getting paid to make CRUD apps is over. Many here will be introduced to AGI faster and sooner. Then it could be phone customer support representatives. It could show up for the face-to-face worker who is now replaced by a screen that can talk to customers (which already arrived yesterday, it's here). It'll appear erratic and not cohesive, unless you zoom out and see the contagion.
---
Rome needed to recognize that the Barbarian hordes had arrived. Pay attention to all the places the invasion has landed. You can pretend like the Vandals are not in your town for a little bit, sure, but eventually they will be knocking on many doors (most likely all doors). We're in a time period of RADICAL transformation. There is no half-assing this conviction. Practicality will not serve us here.
> Are you really going to hire that developer because you need extra manpower to stand up test coverage? No, so as far as that developer is concerned, AGI has arrived for that part of their professional career.
That is exactly what you need in order to make AI useful. Even a baby needs to cry to signal its needs to parents, which are like ASI to it. AI working on a task lacks in 3 domains: start, middle and finish.
AI cannot create its own needs, they belong to the context where it is used. After we set AI to work, it cannot predict the outcomes of its actions unless they pass through your context and return as feedback. In the end, all benefits accumulate in the same context. Not to mention costs and risks - they belong to the context.
The AI is a generalist, context is exactly what it lacks. And context is distributed across people, teams, companies. Context is non-fungible. You can't eat so I get satiated. Context is what drives AI. And testing is the core contextual activity when using AI.
> Let's take someone who once upon a time was responsible for grading papers. As far as that person is concerned, AGI has arrived for their profession
You're talking about TAs. I know TAs. Their jobs have not disappeared. They are not using AI to grade papers.
> Are you really going to hire that developer because you need extra manpower to stand up test coverage?
Yes. Unsupervised AI agents cannot currently replace developers. "Oh we'll get someone to supervise it"—yes, that person's job title is "developer" and they will be doing largely the same job they'd have done 5 years ago.
> The universe of people getting paid to make CRUD apps is over.
Tell that to all the people who get paid to make CRUD apps. Frankly, Airtable has done more to disrupt CRUD apps than AI ever did.
> Rome needed to recognize that the Barbarian hordes had arrived.
IDK what to tell you. All these jobs are still around. You're just fantasizing.
Can you seriously deny that it will take many fewer developers using LLMs to do what was done without LLMs? Many if not most software companies have either greatly slowed hiring, or laid off tranches of developers.
Not all developer jobs will disappear, but there will most certainly be fewer available. Any new grad can tell you how hard it is to find a software engineering job nowadays.
Certainly we cannot just let an AI spin and build software unattended. But what used to take days can now be done in minutes.
> How many people are purchasing AI-generated goods?
Probably a lot. I remember my mom recently showing me an AI-generated book she bought. And pretty much immediately refunded it. Not because it was AI, but because the content was trash.
> Even in the dot-com era, the early internet was already changing media and commerce in fundamental ways.
I agree that AI is overhyped but so was the early web. It was projected to do a lot of things ”soon”, but was not really doing that much 4 years in. I don’t think the newspapers or commerce were really worried about it. The transformation of the business landscape took hold after the crash.
That "factor of 17" comes from an interest rate model that is unrelated to AI.
This is not true. Obviously the underlying effect is real but not nearly to this scale—for instance, neither the CPI nor the S&P500 are even remotely close to 17x higher than they were at the turn of the millennium.
The source is a report written by Julien Garran based on the difference between actual interest rates and an idea of what they should be called the Wicksell spread. There's a summary here https://www.marketwatch.com/story/the-ai-bubble-is-17-times-...
He figured there was a credit bubble like that around the time of the dot com bubble and now but the calculation if purely based on interest rates and the money can go into any assets - property, stocks, crypto etc. It's not AI specific.
He explains it here https://youtu.be/uz2EqmqNNlE
The Wicksell spread seems to have come from Wicksell's proposed 'natural rate of interest' detailed in his 1898 book
https://en.wikipedia.org/wiki/Knut_Wicksell#Interest_and_Pri...
I see, thank you!
> The value created here would have to exceed what the internet was generating
Its precisely why these companies are investing so much, robots combined with AI will be creating that value.
> Robots combined with AI will be creating that value.
Will they? Within what timeframe? Because a bubble economy can't be told to "just hang on a few more years" forever. LLMs are normal technology; they will not suddenly become something they are not. There's no indication that general intelligence is right on the horizon.
If you think they wont then you should short the stocks starting with Nvidia and get rich.
> There's no indication that general intelligence is right on the horizon.
You dont need general intelligence for all the tasks, if a robot can do some of those tasks with limited intelligence cheaper than a humnan, that is all corporations care about.
Almost everyone I hear calling our AI hype machine a bubble aren't claiming AI is a short term fluke. They're saying the marketing doesn't match the reality. The companies don't have the revenue they need. The model performance is hitting the top of the S curve. Essentially, this is the first big wave - but it'll be a while before the sea level rises permanently.
> marketing doesn't match the reality.
true for every marketing ever
It’s not just a marketing stunt, it’s a trillion dollar grift that VCs are going to try to dump off onto the public markets when the reality doesn’t catch up to the hype fast enough
takes a lot of hubris to be sure it's a bubble too.
that's why I always identify the central position of any argument and take it. that way noone can accuse me of hubris
Spoken like a wise man.
You can take a position without being sure about it. e.g. "I'm at 70% that AI is a bubble."
id probably go with 50% actually
The LLM architectures we have now have reached their full potential already, so going further would require something completely different. It isn’t a matter of refining the existing tech, whereas the internet of 1997 is virtually technologically identical to what we have today. The real change has been sociological, not technological.
To make a car analogy; the current LLMs are not the early cars, but the most refined horse drawn carriages. No matter how much money is poured into them, you won’t find the future there.
The current generation of LLM's have convinced me that we already have the compute and the data needed for AGI, we just likely need a new architecture. But I really think such an architecture could be right around the corner. It appears to me like the building blocks are there for it, it would just take someone with the right luck and genius to make it happen.
> The current generation of LLM's have convinced me that we already have the compute and the data needed for AGI, we just likely need a new architecture.
I think this is one of the greatest fallacies surrounding LLMs. This one, and the other one - scaling compute!! The models are plenty fine, what they need is not better models, or more compute, they need better data, or better feedback to keep iterating until they reach the solution.
Take AlphaZero for example, it was a simple convolutional network, not great compared to LLMs, small relative recent models, and yet it beat the best of us at our own game. Why? Because it had unlimited environment access to play games against other variants of itself.
Same for the whole Alpha* family, AlphaStar, AlphaTensor, AlphaCode, AlphaGeometry and so on, trained with copious amounts of interactive feedback could reach top human level or surpass humans in specific domains.
What AI needs is feedback, environments, tools, real world interaction that exposes the limitations in the model and provides immediate help to overcome them. Not unlike human engineers and scientists - take their labs and experiments away and they can't discover shit.
It's also called the ideation-validation loop. AI can ideate, it needs validation from outside. That is why I insist the models are not the bottleneck.
For Alpha Zero, the "better data" was trivial. The environment of board games is extremely simplistic. It just can't be compared to language models.
The problem with language is that there is no know correct answer. Everything is vague, ambiguous and open ended. How would we even implement feedback for that?
So yes, we do need new models.
> The current generation of LLM's have convinced me that we already have the compute and the data needed for AGI, we just likely need a new architecture
This is likely true but not for the reasons you think about. This was arguably true 10 years ago too. A human brain uses 100 watts per day approx and unlike most models out there, the brain is ALWAYS in training mode. It has about 2 petabytes of storage.
In terms of raw capabilities, we have been there for a very long time.
The real challenge is finding the point where we can build something that is AGI level with the stuff we have. Because right now, we might have the compute and data needed for AGI but we might lack the tools needed to build a system that efficient. It's like a little dog trying to enter a fenced house, the closest path topologically between the dog and the house might not be accessible for that dog at that point because its current capabilities (short legs, inability to jump high or push through the fence standing in between) so while it is further topologically, a longer path topologically might be the closest path to reach the house.
In case it's not obvious, AGI is the house, we are the little dog and the fence represent current challenges to build AGI.
The notion that the brain uses less energy than an incandescent lightbulb and can store less data than YouTube does not mean we have had the compute and data needed to make AGI "for a very long time".
The human brain is not a 20-watt computer ("100 watts per day" is not right) that learns from scratch on 2 petabytes of data. State manipulations performed in the brain can be more efficient than what we do in silicon. More importantly, its internal workings are the result of billions of years of evolution, and continue to change over the course of our lives. The learning a human does over its lifetime is assisted greatly by the reality of the physical body and the ability to interact with the real world to the extent that our body allows. Even then, we do not learn from scratch. We go through a curriculum that has been refined over millennia, building on knowledge and skills that were cultivated by our ancestors.
An upper bound of compute needed to develop AGI that we can take from the human brain is not 20 watts and 2 petabytes of data, it is 4 billion years of evolution in a big and complex environment at molecular-level fidelity. Finding a tighter upper bound is left as an exercise for the reader.
> it is 4 billion years of evolution in a big and complex environment at molecular-level fidelity. Finding a tighter upper bound is left as an exercise for the reader.
You have great points there and I agree. Only issue I take with your remark above. Surely, by your own definition, this is not true. Evolution by natural selection is not a deterministic process so 4 billion years is just one of many possible periods of time needed but not necessarily the longest or the shortest.
Also, re "The human brain is not a 20-watt computer ("100 watts per day" is not right)", I was merely saying that there exist an intelligence that consumes 20 watts per day. So it is possible to run an intelligence on that much energy per day. This and the compute bit do not refer to the training costs but to the running costs after all, it will be useless to hit AGI if we do not have enough energy or compute to run it for longer than half a millisecond or the means to increase the running time.
Obviously, the path to design and train AGI is going to take much more than that just like the human brain did but given that the path to the emergence of the human brain wasn't the most efficient given the inherent randomness in evolution natural selection there is no need to pretend that all the circumstances around the development of the human brain apply to us as our process isn't random at all nor is it parallel at a global scale.
> Evolution by natural selection is not a deterministic process so 4 billion years is just one of many possible periods of time needed but not necessarily the longest or the shortest.
That's why I say that is an upper bound - we know that it _has_ happened under those circumstances, so the minimum time needed is not more than that. If we reran the simulation it could indeed very well be much faster.
I agree that 20 watts can be enough to support intelligence and if we can figure out how to get there, it will take us much less time than a billion years. I also think that on the compute side for developing the AGI we should count all the PhD brains churning away at it right now :)
"watts per day" is just not a sensible metric. watts already has the time component built in. 20 watts is a rate of energy usage over time.
Dial-up modems reached their full 56kbps potential in 1997, and going further required something completely different. It happened naturally to satisfy demand, and was done by many of the same companies and people; the change was technological, not sociological.
I think we're probably still far from the full potential of LLMs, but I don't see any obstacles to developing and switching to something better.
I don't think that comparison works very well at all.
We had plenty of options for better technologies both available and in planning, 56k modems were just the cost effective/lowest common denominator of their era.
It's not nearly as clear that we have some sort of proven, workable ideas for where to go beyond LLMs.
> Dial-up modems reached their full 56kbps potential in 1997
That's simply not true. Modems were basically the same tech in the dsl era, and using light instead of electricity is a very gradual refinement.
> we're probably still far from the full potential of LLMs
Then how come the returns are so extremely diminishing?
> I don't see any obstacles to developing and switching to something better.
The obstacle is that it needs to be invented. There was nothing stopping newton from discovering relativity either. We simply have no idea what the road forward even looks like.
> The LLM architectures we have now have reached their full potential already.
How do we know that?
what we can say right now is that we've hit the point of diminishing returns and the only way we're going to get signicantly more capable models is through a technological advance that we cannot forsee (and that may not come for decades if it ever comes)
Exactly. You're absolutely right to focus on that.
You could see some potential modifications. Already some are multimodal. You'd probably want something to change the weights as time goes on so they can learn. It might be more steam engines needing to be converted to petrol engines.
> If you told someone in 1995 that within 25 years [...] most people would find that hard to believe.
That's not how I remember it (but I was just a kid so I might be misremembering?)
As I remember (and what I gather from media from the era) late 80s/early 90s were hyper optimistic about tech. So much so that I distinctly remember a ¿german? TV show when I was a kid where they had what amounts to modern smartphones, and we all assumed that was right around the corner. If anything, it took too damn long.
Were adults outside my household not as optimistic about tech progress?
To your point, AT&T's "You Will" commercials started airing in 1993 and present both an optimistic and fairly accurate view of what the future would look like.
https://www.youtube.com/watch?v=RvZ-667CEdo
I didn't know about these ads, thanks for sharing! Can't imagine how people reacted to that when they aired — the things they described sound so "normal" today, I wonder if it was seen as far fetched, crazy or actually expected.
I was in my late teens at the time. My memory is that I felt like the tech was definitely going happen in some form, but I rolled my eyes heavily at the idea that AT&T was going to be the company to do make it happen.
If you’re unfamiliar, the phone connectivity situation in the 80s and 90s was messy and piecemeal. AT&T had been broken up in 1982 (see https://www.historyfactory.com/insights/this-month-in-busine...), and most people had a local phone provider and AT&T was the default long-distance provider. MCI and Sprint were becoming real competition for AT&T at the time of these commercials.
Anyway, in 1993 AT&T was still the crusty old monopoly on most people’s minds, and the idea that they were going to be the company to bring any of these ideas to the market was laughable. So the commercials were basically an image play. The only thing most people bought from AT&T was long distance service, and the main threat was customers leaving for MCI and Sprint. The ads memorable for sure, but I don’t think they blew anyone’s mind or made anyone stay with AT&T.
We’re the same age, and I had exactly the same reaction.
AT&T and the baby bells were widely loathed (man I hated Ameritech…), so the idea they would extend their tentacles in this way was the main thing I reacted to. The technology seemed straightforwardly likely with Dennard scaling in full swing.
I thought it would be banks that owned the customer relationship, not telcos or Apple (or non-existent Google), but the tech was just… assume miniaturization’s plateau isn’t coming for a few decades.
Still pretty iconic/memorable, though!
Thanks for that context. Was your expectation/prediction about that tech that it was going to happen in say 2000s or sooner/later?
In these commercials, it wasn't the technology itself but the ease of access and visualized integration of these technologies into the commoners' everyday lives that was the new idea.
Wow, that genuinely gave me goosebumps. It is incredible to live in a time where so much of that hopeful optimism came to pass.
Same here. It's wild that every single thing in that video came to pass, and more. Within our own lifetimes.
Still no flying cars, though.
Indeed, AI now is what people in the 1980s thought computers would be doing in 2000.
Except people thought it would get basic facts right.
We can't decide whether to take a vaccine even when we are dying left and right. And we have brains, not chips inside.
Or hell, as Neil deGrasse Tyson said in a video, just put 2 lines with different arrows at the ends then our brains can't even tell if they're the same size!
https://en.wikipedia.org/wiki/Müller-Lyer_illusion
That’s how I remember it too. The video is from 1999, during the height of the dot-com bubble. These experts are predicting that within 10 years the internet will be on your phone, and that people will be using their phones as credit cards and the phone company would manage the transaction, the prediction actually comes pretty close to the prediction made by bitcoin enthusiasts.
https://bsky.app/profile/ruv.is/post/3liyszqszds22
Note that this is the state TV broadcasting this in their main news program. The most popular daily show in Iceland.
Still waiting on my flying car.
To be fair, that has been a Sci-Fi trope for at least 130 years and predates the invention of the car itself (e.g. personal wings/flying horse -> flying ship -> personal balloon -> flying automobile). So countless generations have been waiting for that :)
Might not be waiting for long.
There's no way I'm trusting the current driving cohort with a third dimension. If we get flying cars and they aren't completely autonomous, I am moving to the sticks.
Self-flying cars? I wonder if it's actually easier to have autonomous vehicles operating in 3D than in "2D".
The vast majority of the dot-com comparison that I personally see are economic, not technological. People (or at least the ones I see) are claiming that the bubble mechanics of e.g. circular trading and over-investments are similar to the dot-com bubble, not that the AI technology is somehow similar the internet (it obviously isn’t). And to that extent we are in the year 1999 not 1995.
When this article are claiming both sides of the debate, I believe only one of them are real (the ones hyping up the technology). While there are people like me who are pessimistic about the technology, we are not in any position of power, and our opinion on the matter is basically a side noise. I think a much more common (among people with any say in the future of this technology) is the believe that this technology is not yet at a point which warrants all this investment. There were people who said that about the internet in 1999, and they were proven 100% correct in the months that followed.
Agreed. It would probably be better to keep improving AI before investing that much into infrastructure.
I'm getting ai fatigue. It's ok to rewrite quick emails that i'm having brain farts on but anything deep it just sucks. I certainly can't see paying for it.
Weird because AI has been solving hard problems for me. Even finding solutions that I couldn’t find myself. Ie. sometimes my brain cant wrap around a problem, I throw it to AI and it perfectly solves it.
I pay for chatgpt plus and github copilot.
It is weird that AI is solving hard problems for you. I can't get it to do the most basic things consistently, most of the time it's just pure garbage. I'd never pay for "AI" because it wastes more of my time than it saves. But I've never had a problem wrapping my head around a problem, I solve problems.
I'm curious what kind of problem your "brain cant wrap around", but the AI could.
Is that a "hard problem" though? Really?
Yes. To me, it is. Sometimes queries I give it are 100-200 lines long. Sure, I can solve it eventually but getting an "instant" answer that is usually correct? Absolutely priceless.
It's pretty common for me to spend a day being stuck on a gnarly problem in the past. Most developers have. Now I'd say that's extremely rare. Either an LLM will solve it outright quickly or I get enough clues from an LLM to solve it efficiently.
You might be robbing yourself of the opportunity to learn SQL for real by short-cutting to a solution that might not even be correct one.
I've tried using LLMs for SQL and it fails at exactly that: complexity. Sure it'll get the basic queries right, but throw in anything that's not standard every day SQL into it and it'll give you solutions that are not great really confidently.
If you don't know SQL enough to figure out these issues in the first place, you don't know if the solutions the LLM provides are actually good or not. That's a real bad place to be in.
Usually the term, "hard problem", is reserved for problems that require novel solutions
Have you ever read Zen and the Art of Motorcycle Maintenance? One of the first examples in that book is how when you are disassembling a motorcycle any one bolt is trivial until one is stuck. Then it becomes your entire world for a while as you try to solve this problem and the solution can range from trivial to amazingly complex.
You are using the term “hard problem” to mean something like solving P = NP. But in reality as soon as you step outside of your area of expertise most problems will be hard for you. I will give you some examples of things you might find to be hard problems (without knowing your background):
- what is the correct way to frame a door into a structural exterior wall of a house with 10 foot ceilings that minimized heat transfer and is code compliant.
- what is the correct torque spec and sequence for a Briggs and Stratton single cylinder 500 cc motor.
- how to correctly identify a vintage Stanley hand plane (there were nearly two dozen generations of them, some with a dozen different types), and how to compare them and assess their value.
- how to repair a cracked piece of structural plastic. This one was really interesting for me because I came up with about 5 approaches and tried two of them before asking an LLM and it quickly explained to me why none of the solutions I came up with would work with that specific type of plastic (HDPE is not something you can glue with most types of resins or epoxies and it turns out plastic welding is the main and best solution). What it came up with was more cost efficient, easier, and quicker than anything I thought up.
- explaining why mixing felt, rust, and CA glue caused an exothermal reaction.
- find obscure local programs designed to financially help first time home buyers and analyze their eligibility criteria.
In all cases I was able to verify the solutions. In all cases I was not an expert on the subject and in all cases for me these problems presented serious difficulty so you might colloquially refer to them as hard problems.
It is not. It’s relative to the subject.
In this case, the original author stated that AI only good for rewriting emails. I showed a much harder problem that AI is able to help me with. So clearly, my problem can be reasonably described as “hard” relative to rewriting emails.
If you have 200 line SQL queries you have a whole other kind of problem.
not unless you are working on todo apps.
TODO: refactor the schema design.
What happens when these "AI" companies start charging you what it really costs to run the "AI"? You'd very likely balk at it and have to learn SQL yourself. Enjoy it while it lasts, I guess?
Problem with this is people will accept tech debt and slow query's so long as the LLM can make sense of it (allegedly!).
So the craft is lost. Making that optimised query or simplifying the solution space.
No one will ask "should it be relational even?" if the LLM can spit out sql then move on to next problem.
So why not ask the LLM if it should be relational and provide the pros and cons?
Anyway, I'm sure people have asked if we should be programming in C rather than Assembly to preserve the craft.
Surely you understand the difference between not knowing how to do anything by yourself and only knowing how to use high-level languages?
That is like using the LLM like a book. Sure do that! But human still needs to understand and make the decisions.
I work with some very complex queries (that I didn't write), and yeah, AI is an absolute lifesaver, especially in troubleshooting situations. What used to take me hours now takes me minutes.
In my case, Learning new stuff is one place I see AI playing major role. Especially the academic research which is hard to start if you are newbie but with AI I can start my research, read more papers with better clarity.
Which model are you using?
Sounds like you're not capable of using AI correctly, user error.
Sorry, I'm not taking a comment like this from a 2-hour old account seriously. You don't know me at all.
"It can't be that stupid, you must be prompting it wrong!"
Sigh.
Can you give some examples??
Calculate the return on investment for a solar installation of a specified size on a specified property based on the current dynamic prices of all of the panels, batteries, inverter, and balance of system components, the current zoning and electrical code, the current cost of capital, the average insolation and weather taking into account likely changes in weather in the future as weather instability increases due to more global increase of temperature, the chosen installation method and angle, and the optimal angle of the solar panels if adjusted monthly or quarterly. Now do a Manual J calculation to determine the correct size of heat pump in each section of that property, taking into account number of occupants, insulation level, etc.
ChatGPT is currently the best solar calculator on the publicly accessible internet and it's not even close. It'll give you the internal rate of return, it'll ask all the relevant questions, find you all the discounts you can take in taxes and incentives, determine whether you should pay the additional permitting and inspection cost for net metering or just go local usage with batteries, size the batteries for you, and find some candidate electricians to do the actual installation once you acquire the equipment.
Edit: My guess is that it'd cost several thousand dollars to hire someone to do this for you, and it'll save you probably in the $10k-$30k range on the final outcomes, depending on the size of system.
My God, the first example is having an AI do math, then he says "Well I trust it to a standard deviation"
So it's literally the same as googling "what's the ballpark solar installation cost for X in Y area" unbelievable, and people pay $20+ per month for this
$200 :)
Any way to tell if the convincing final numbers it told you are real or halucinated ?
Solve the same task with ChatGPT, Gemini and Claude. If they agree, you can be reasonably sure.
I'm not opposed to experimenting, but that's a a recipe for false confidence in a final decision.
Where they agree it shows the data supports that answer - not necessarily that it is true, where they disagree it shows you need to hedge. That's useful.
This is so wrong!
e.g., if you had a heart condition, you can't just poll three LLMs and be "reasonably sure" you've properly diagnosed the ailment.
I checked them carefully myself with various other tools. It was using python to do the math so I trust it to a single standard deviation at least.
Standard deviation of what
I'm lost too. Financials are technology agnostic.
They probably meant that they could read (and trace) the logic in Python for correctness.
I wouldn't publish it as statistically significant but it's within the error bounds for a real human accomplishing the same task, to reword.
I see your edit.
I would recommend spending that "couple thousand" for quote(s). It's a second opinion from someone who hopefully has high volume in your local market. And your downside could be the entire system plus remediation, fines, etc.
To be clear, I'm not opposed to experimenting, but I wouldn't rely on this. Appreciate your comment for the discussion.
No I'm not relying on it in the sense of going out and running the entire project through it, but as an accurate screener for whether it's worth doing, there's nothing comparable available.
>Weird because AI has been solving hard problems for me.
Examples or it didn't happen.
Well deep/hard is different I guess; I use it, day and night, for things I find boring. Boilerplate coding (which now is basically everything that's not pure business logic / logic / etc), corporate docs, reports etc. Everything I don't want to do is done by AI now. It's great. Outside work I use it for absolutely nothing though; I am writing a book, framework and database; that's all manual work (and I don't AI is good at any of those (yet)).
As an LLM-skeptic who got a Claude subscription, the free models are both much dumber and configured for low latency and short dumb replies.
No it won’t replace my job this year or the next, but what Sonnet 4.5 and GPT 5 can do compared to e.g. Gemini Flash 2.5 is incredible. They for sure have their limits and do hallucinate quite a bit once the context they are holding gets messy enough but with careful guidance and context resets you can get some very serious work done with them.
I will give you an example of what it can’t do and what it can: I am working on a complicated financial library in Python that requires understanding nuanced parts of tax law. Best in class LLM cannot correctly write the library code because the core algorithm is just not intuitive. But it can:
1. Update all invocations of the library when I add non-optional parameters that in most cases have static values. This includes updating over 100 lengthy automated tests.
2. Refactor the library to be more streamlined and robust to use. In my case I was using dataclasses as the base interface into and out of it and it helped me split one set of classes into three: input, intermediate, and output while fully preserving functionality. This was a pattern it suggested after a changing requirement made the original interface not make nearly as much sense.
3. Point me to where the root cause of failing unit tests was after I changed the code.
4. Suggest and implement a suite of new automated tests (though its performance tests were useless enough for me to toss out in the end).
5. Create a mock external API for me to use based on available documentation from a vendor so I could work against something while the vendor contract is being negotiated.
6. Create comprehensive documentation on library use with examples of edge cases based on code and comments in the code. Also generate solid docstrings for every function and method where I didn’t have one.
7. Research thorny edge cases and compare my solutions to commercial ones.
8. Act as a rubber ducky when I had to make architectural decisions to help me choose the best option.
It did all of the above without errors or hallucinations. And it’s not that I am incapable of doing any of it, but it would have taken me longer and would have tested my patience when it comes to most of it. Manipulating boilerplate or documenting the semantic meaning between a dozen new parameters that control edge case behavior only relevant to very specific situations is not my favorite thing to do but an LLM does a great job of it.
I do wish LLMs were better than they are because for as much as the above worked well for me, I have also seen it do some really dumb stuff. But they already are way too good compared to what they should be able to do. Here is a short list of other things I had tried with them that isn’t code related that has worked incredibly well:
- explaining pop culture phenomenon. For example I had never understood why Dr Who fans take a goofy campy show aimed in my opinion at 12 year olds as seriously as if it was War and Peace. An LLM let me ask all the dumb questions I had about it in a way that explained it well.
- have a theological discussion on the problem of good and evil as well as the underpinnings of Christian and Judaic mythology.
- analyze in depth my music tastes in rock and roll and help fill in the gaps in terms of its evolution. It actually helped me identify why I like the music I like despite my tastes spanning a ton of genres, and specifically when it comes to rock, created one of the best and most well curated playlists I had ever seen. This is high praise for me since I pride myself on creating really good thematic playlists.
- help answer my questions about woodworking and vintage tool identification and restoration. This stuff would have taken ages to research on forums and the answers would still be filled with purism and biased opinions. The LLM was able to cut through the bullshit with some clever prompting (asking it to act as two competing master craftsmen).
- act as a writing critic. I occasionally like to write essays on random subjects. I would never trust an LLM to write an original essay for me but I do trust it to tell me when I am using repetitive language, when pacing and transitions are off, and crucially how to improve my writing style to take it from B level college student to what I consider to be close to professional writer in a variety of styles.
Again I want to emphasize that I am still very much on the side of there being a marketing and investment bubble and that what LLMs can do being way overhyped. But at the same time over the last few months I have been able to do all of the above just out of curiosity (the first coding example aside). These are things I would have never had the time or energy to get into otherwise.
You seem very thoughtful and careful about all this, but I wonder how you feel about the emergence of these abilities in just 3 years of development? What do you anticipate it will be capable of in the next 3 years?
With no disrespect I think you are about 6-12 months behind SOTA here, the majority of recent advances have come from long running task horizons. I would recommend to you try some kind of IDE integration or CLI tool, it feels a bit unnatural at first but once you adapt your style a bit, it is transformational. A lot of context sticking issues get solved on their own.
Oh I am very much catching up. I am suing Claude Code primarily, and also have been playing a bit with all the latest API goodies from OpenAI and Anthropic, like custom tools, memory use, creating my own continuous compaction algorithm for a specific workflow I tried. There is a lot happening here very fast.
One thing that struck me: models are all trained starting 1-2 years ago. I think the training cutoff for Sonnet 4.5 is like May 2024. So I can only imagine with is being trained and tested currently. And also these models are just so ahead of things like Qwen and Llama for the types of semi-complex non-coding tasks I have tried (like interpreting my calendar events), that it isn’t even close.
[dead]
“But the fact that some geniuses were laughed at does not imply that all who are laughed at are geniuses. They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown.”
Because some notable people dismissed things that wound up having profound effect on the world, it does not mean that everything dismissed will have a profound effect.
We could just as easily be "peak Laserdisc" as "dial-up internet".
I was happy to come into this thread and see I was not the first person for whom that quote came to mind. The dial-up Internet comparison implicitly argues for a particular outcome of current AI as a technology, but doesn't actually support that argument.
There's another presumably unintended aspect of the comparison that seems worth considering. The Internet in 2025 is certainly vastly more successful and impactful than the Internet in the mid-90s. But dial-up itself as a technology for accessing the Internet was as much of a dead-end as Laserdisc was for watching movies at home.
Whether or not AI has a similar trajectory as the Internet is separate from the question of whether the current implementation has an actual future. It seems reasonable to me that in the future we're enjoying the benefits of AI while laughing thinking back to the 2025 approach of just throwing more GPUs at the problem in the same way we look back now and get a chuckle out of the idea of "shotgun modems" as the future.
While I mostly agree with the article's premise (that AI will cause more software development to happen, not less) I disagree with two parts:
1. the opening premise comparing AI to dial-up internet; basically everyone knew the internet would be revolutionary long before 1995. Being able to talk to people halfway across the world on a BBS? Sending a message to your family on the other side of the country and them receiving it instantly? Yeah, it was pretty obvious this was transformative. The Krugman quote is an extreme, notable outlier, and it gets thrown out around literally every new technology, from blockchain to VR headsets to 3DTVs, so just like, don't use it please.
2. the closing thesis of
> Consider the restaurant owner from earlier who uses AI to create custom inventory software that is useful only for them. They won’t call themselves a software engineer.
The idea that restaurant owners will be writing inventory software might make sense if the only challenge of creating custom inventory software, or any custom software, was writing the code... but it isn't. Software projects don't fail because people didn't write enough code.
Before I got my first full time software engineering gig (I had worked part time briefly years prior) I was working full time as a carpenter. We were paying for an expensive online work order system. Having some previous experience writing software for music in college and a couple /brief/ LAMP stack freelance jobs after college I decided to try to write my own work order system. It took me like a month and it would never have never scaled, was really ugly, and had the absolute minimum number of features. I could never had accepted money from someone to use it but it did what we needed and we ran with it for several years after that.
I was only able to do this because I had some prior programming experience but I would imagine that if AI coding tools get a bit better they would enable a larger cohort of people to build a personal tool like I did.
I don't think his quote is that extreme and it was definitely not obvious to most people. A common thing you heard even around 95 was "I've tried internet but it was nothing special".
> basically everyone knew the internet would be revolutionary long before 1995. Being able to talk to people halfway across the world on a BBS? Sending a message to your family on the other side of the country and them receiving it instantly? Yeah, it was pretty obvious this was transformative.
That sounds pretty similar to long-distance phone calls? (which I'm sure was transformative in its own way, but not on nearly the same scale as the internet)
Do we actually know how transformative the general population of 1995 thought the internet would or wouldn't be?
In 1995 in France we had the minitel already (like really a lot of people had one) and it was pretty incredible, but we were longing for something prettier, cheaper, snappier and more point to point (like the chat apps or emails).
As soon as the internet arrived, a bit late for us (I'd say 1999 maybe) due to the minitel's "good enough" nature, it just became instantly obvious, everyone wanted it. The general population was raving mad to get an email address, I never heard anyone criticize the internet like I criticize the fake "AI" stuff now.
>> Consider the restaurant owner from earlier who uses AI to create custom inventory software that is useful only for them. They won’t call themselves a software engineer.
I have a suspicion this is LLM text, sounds corny. There are dozens open source solutions, just look one up.
Great analysis but one thing overlooked is that current gen advanced AI could in five or ten years (or less) be run from the smartphone or desktop, which could negate all the capex from the hyperscalers and also Nvidia, which presents a massive target for competitors right now. The self same AI revolution we’re seeing created right now could take itself down if AI tooling becomes widespread.
If this happen, everyone's computer will contain one Nvidia GPU.
Not really. Apple is a very strong competitor here.
It's more like the Segway era when people with huge stakes in Segway tried to convince the world we were about to rebuild entire cities around the new model.
I recall the unit economics making sense for all these other industries and bubbles (short of maybe tulips, which you could plant…) . Sure there were over-valuation bubbles because of speculatory demand, but right now the assumption seems to be “first to AGI wins” but that… may not happen.
The key variable for me in this house of cards is how long folks will wait before they need to see their money again, and whether these companies will go in the right direction long enough given these valuations to get to AGI. Not guaranteed and in the meantime society will need to play ball (also not a guarantee)
The article seems well researched, has some good data, and is generally interesting. It's completely irrelevant to the reality of the situation we are currently in with LLMs.
It's falling into the trap of assuming we're going to get to the science fiction abilities of AI with the current software architectures, and within a few years, as long as enough money is thrown at the problem.
All I can say for certain is that all the previous financial instruments that have been jumped on to drive economic growth have eventually crashed. The dot com bubble, credit instruments leading to the global financial crisis, the crypto boom, the current housing markets.
The current investments around AI that we're all agog at are just another large scale instrument for wealth generation. It's not about the technology. Just like VR and BioTech wasn't about the technology.
That isn't to say the technology outcomes aren't useful and amazing, they are just independant of the money. Yes, there are Trillions (a number so large I can't quite comprehend it to be honest) being focused into AI. No, that doesn't mean we will get incomprehensible advancements out the other end.
AGI isn't happening this round folks. Can hallucinations even be solved this round? Trillions of dollars to stop computers lying to us. Most people where I work don't even realise hallucinations are a thing. How about a Trillion dollars so Karen or John stop dismissing different viewpoints because a chat bot says something contradictory, and actually listen? Now that would be worth a Trillion dollars.
Imagine a world where people could listen to others outside of their bubble. Instead they're being given tools that re-inforce the bubble.
Indeed, this could be AI's fusion energy era, or AI's VR era, or even AI's FTL travel era.
People keep comparing the AI boom to the Dotcom bubble. They’re wrong. Others point to the Railway Mania of the 1840s — closer, but still not quite right.
The real parallel is Canal Mania — Britain’s late-18th-century frenzy to dig waterways everywhere. Investors thought canals were the future of transport. They were, but only briefly.
Today’s AI runs on GPUs — chips built for rendering video games, not thinking machines. Adapting them for AI is about as sensible as adapting a boat to travel across land. Sure, it moves — but not quickly, not cheaply, and certainly not far.
It works for now, but the economics are brutal. Each new model devours exponentially more power, silicon, and capital. It just doesn't scale.
The real revolution will come with new, hardware built for the job (that hasn't been invented yet) — thousands of times faster and more efficient. When that happens, today’s GPU farms will look like quaint relics of an awkward, transitional age: grand, expensive, and obsolete almost overnight.
I think specialized hardware will emerge for specific proven workloads (transformer inference, for example), but GPUs won't become obsolete. They'll remain the experimentation platform for new architectures. You need flexibility to discover what's worth building custom silicon for.
Think 3D printers versus injection molds: you prototype with flexibility, then mass-produce with purpose-built tooling. We've seen this pattern before too. CPUs didn't vanish when GPUs arrived for graphics. The canal analogy assumes wholesale replacement. Reality is likely more boring: specialization emerges and flexibility survives.
Sure, but your R&D infrastructure isn't going to be 1.5 trillion dollars.
> Today’s AI runs on GPUs — chips built for rendering video games, not thinking machines. Adapting them for AI is about as sensible as adapting a boat to travel across land.
A GPU is fundamentally just a chip for matrix operations, and that's good for graphics but also for "thinking machines" as we currently have them. I don't think it's like a boat traveling on land at all.
Another definition: A modern GPU a general purpose computer that can make parallelized and efficient computations. It's optimized to run limited number of operations but on large number of data points.
This happens to be useful both for graphics (same "program" running on on huge number of pixels/vertices) and neural networks (same neural operations on huge number of inputs/activations)
Nvidia’s enterprise GPUs having nothing to do with graphics anymore except for the name.
GPUs are massively parallel, sure, but they still have a terrible memory architecture and are difficult to program (and are still massively memory constrained). It's only NVidia's development in cuda that made it even feasible to create decent ML models on GPUs.
I think it'll be a combination of hardware of course, but also better software - surely there is a better way of doing this (like our brains do) which will eventually require less power
Recently, in my city, the garbage trucks started to come equipped with a device I call "The Claw" (think Toy Story). The truck drives to your curb where your bin is waiting, and then The Claw extends, grasps the bin, lifts it into the air and empties the contents into the truck before setting it down again.
The Claw allows a garbage truck to be crewed by one man where it would have needed two or three before, and to collect garbage much faster than when the bins were emptied by hand. We don't know what the economics of such automation of (physical) garbage collection portend in the long term, but what we do know is that sanitation workers are being put out of work. "Just upskill," you might say, but until Claw-equipped trucks started appearing on the streets there was no need to upskill, and now that they're here the displaced sanitation workers may be in jeopardy of being unable to afford to feed their families, let alone find and train in some new marketable skill.
So no, we're in the The Claw era of AI, when business finds a new way to funge labor with capital, devaluing certain kinds of labor to zero with no way out for those who traded in such labor. The long-term implications of this development are unclear, but the short-term ones are: more money for the owner class, and some people are out on their ass without a safety net because this is Goddamn America and we don't brook that sort of commie nonsense here.
FYI, this kind of garbage truck has been around for >50 years [0], so any wide-scale impact on employment from this technology has likely already settled out.
The waste collection companies in my area don't use them because it's rural and the bins aren't standardized. The side loaders don't work for all use cases of garbage trucks.
[0] https://en.wikipedia.org/wiki/Garbage_truck
>In 1969, the city of Scottsdale, Arizona introduced the world's first automated side loader. The new truck could collect 300 gallon containers in 30 second cycles, without the driver exiting the cab
Every few years I find myself thinking, "Wow...the latest tech is amazing! We were in the stone ages just a few years ago."
I don't expect that to cease in my lifetime.
I feel like I've been in such a phase for a good 10-15 years now. I miss looking back and marvel at the technological progress we've made, and eagerly anticipating the next exciting thing. For a while now, I look back wondering why so few people seem bothered by the fact that technology becomes worse every year, with fewer features, lower quality, and more stupidity; and new developments are no longer about exciting improvements, but dreading what new thing that used to work perfectly fine is broken now.
For a bit, I thought science and industry were finally starting to see the problem with our quality degradation and tech regression. Instead, the current hype cycle is all about settling for even crappier quality and lower reliability.
how much does the correction here hew to making an AI model just look like standardized API calls with predictable responses? If you took away all the costs (data centers, water consumption, money, etc) I still wouldn't use an LLM as a first choice because it's wrong enough of the time to make it useless -- I have to verify everything it says, which is how I would have approached a task in the first place. If we put that analogy into manufacturing, it's "I have to QA everything off of the line _without exception_ and I get frequent material waste"
If you make the context small enough, we're back at /api/create /api/read /api/update /api/delete; or, if you're old-school, a basic function
The only problem is, similarity with dotcom might only go thus far. For example, dotcom bubble itself might not have a similar thing in the past at that time. This is because the overall world context is different and interaction of social, political and economic forces is different.
So, when people say something about future, they are looking into the past to draw some projections or similar trends, but they may be missing the change in the full context. The considered factors of demand and automation might be too few to understand the implications. What about political, social and economic landscape? The systems are not so much insulated to study using just a few factors.
There’s a big difference between the fibre infrastructure left by the dotcom crash, and the GPUs that AI firms will leave behind.
In the dial-up era, the industry was young, there were no established players, it was all a big green field.
The situation is far from similar now. Now there's an app for everything and you must use all of them to function, which is both great and horrible.
From my experience, current generation of AI is unreliable and so cannot be trusted. It makes non-obvious mistakes and often sends you off on tangents, which consumes energy and leads to confusion.
It's an opinion I've built up over time from using AI extensively. I would have expected my opinion to improve after 3 years, but it hasn't.
People tend to equate this to the railroad boom when saying that infrastructure spending will yield durable returns into the future no matter what.
When the railroad bubble popped we had railroads. Metal and sticks, and probably more importantly, rights-of-way.
If this is a bubble, and it pops, basically all the money will have been spent on Nvidia GPUs that depreciate to 0 over 4 years. All this GPU spending will need to be done again, every 4 years.
Hopefully we at least get some nuclear power plants out of this.
Yeah, the short-lived GPU deprecation cycle does feel very relevant here.
I'm still a fan of the railroad comparisons though for a few additional reasons:
1. The environmental impact of the railroad buildout was almost incomprehensibly large (though back in the 1800s people weren't really thinking about that at all.)
2. A lot of people lost their shirts investing in railroads! There were several bubbly crashes. A huge amount of money was thrown away.
3. There was plenty of wasted effort too. It was common for competing railroads to build out rails that served the same route within miles of each other. One of them might go bust and that infrastructure would be wasted.
The boom might not last long enough for western countries to pull heads out of their collective asses and ramp up production of nuclear plants.
It takes China 5 years now, but they've been ramping up for more than 20 years.
Imagine the progress we could have made on climate change if this money had been funneled into that, instead of making some GPU manufacturers obscenely wealthy.
Throwing away the future for "AI" slop.
Yeah it’s infuriating to think about.
What percentage of data centre build costs are the GPUs vs power stations, water cooling plants, buildings, roads, network, racks, batteries, power systems, etc
A bunch of the money is being spent on data centers and their associated cooling and power systems and on the power plants and infrastructure. Those should have much longer depreciation schedules.
There's a lot more to infrastructure spending than GPUs. Companies are building data centers, cooling systems, power plants (including nuclear), laying cables under oceans, launching satellites. Bubble or not, all of this will continue to be useful for decades in the future.
Heck if nothing else all the new capacity being created today may translate to ~zero cost storage, CPU/GPU compute and networking available to startups in the future if the bubble bursts, and that itself may lead to a new software revolution. Just think of how many good ideas are held back today because deploying them at scale is too expensive.
> including nuclear
Note that these are just power purchase agreements. It's not nothing, but it's a long ways away from building nuclear.
The recycling industry will boom. From what demand you ask? We'll find out soon enough.
>> basically all the money will have been spent on Nvidia GPUs that depreciate to 0 over 4 years
I agree the depreciation schedule always seems like a real risk to the whole financial assumptions these companies/investors make, but a question I've wondered: - Will there be an unexpected opportunity when all these "useless" GPUs are put out to pasture? It just seems like saying a factory will be useless because nobody wants to buy an IBM mainframe, but an innovative company can repurpose a non-zero part of that infrastructure for another use case.
Railroads need repair too? Not sure if it's every 4 years. Also, the trains I take to/from work are super slow because there is no money to upgrade.
I think we may not upgrade every 4 years, but instead upgrade when the AI models are not meeting our needs AND we have the funding & political will to do the upgrade.
Perhaps the singularity is just a sigmoid with the top of the curve being the level of capex the economy can withstand.
For what it's worth they cost a lot less than highways to maintain. Something like the 101 in the Bay Area costs about $40,000 per lane-mile per year, or about $240,000.
Trains are closer to $50-100,000 per mile per year.
If there's no money for the work it's a prioritization decision.
I think the hardware infrastructure may be obsolete but at the moment we are still just beginning to figure out how to use AI. So the knowledge will be the important thing that’s left after the bubble. The current infrastructure will probably be as obsolete as dial up infrastructure.
This is precisely why the AI bubble is so much worse than previous bubbles: the main capital asset that the bubble is acquiring is going to depreciate before the bubble's participants can ever turn a profit. Regardless of what AI's future capabilities are going to be, it's physically impossible for any of these companies to become profitable before the GPUs that they have already purchased are either obsolete or burnt out from running under heavy load.
What are the disadvantages of AI?
The author didn't mention them.
AI companies robbed so much data from the Internet free and without permission.
Sacrificing the interests of owners of websites.
It's not sustainable.
It's impossible for AI to go far.
I sometimes wonder, in a world where the data becomes overwhelmingly AI-generated, if AI starts feeding on itself, a copy of a copy of a copy.
We’re already seeing this sort of well poisoning occur.
There are some gross approximations in the comparison. Oversized fibre optics networks laid out in the late 90s were used for years and may even be in part still used today; today's servers and GPUs will be obsolete in 3 to 5 years, and not worth their weight in scrap metal in 10.
The part about Jevons' paradox is interesting though.
There’s a big difference between the fibre infrastructure left by the dotcom crash, and the GPUs that AI firms will leave behind
One thing the analysis for textiles vs cars misses I the complexity of the supply chain and the raw materials / components that need to be procured to make the end product. Steel/textiles have simple supply chains and they went through a boom/bust cycle as the demand plateaued. But cars on the other hand will not go through the same pattern - there are too many logistical things that need to line up and the trust factor in each of those steps as well as the end product is quite high.
Software is similar to cars - the individual components that need to be properly procured and put together is very complex and trust will be important - will you trust that you as a restaurant owner vibe coded your payment stack properly or will you just drop in the 3 lines to integrate with Stripe? I think most of the non-tech business owners will do the latter.
It's clear that AI is useful. It's not yet clear how useful. Hype has always obscured real value, and nobody knows the real value until the hype cycle completes.
What is clear, is that we have strapped a rocket to our asses, fueled with cash and speculation. The rocket is going so fast we don't know where we're going to land, or if we'll land softly, or in a very large crater. The past few decades have examples of craters. Where there are potential profits, there are people who don't mind crashing the economy to get them.
I don't understand why we're allowing this rocket to begin with. Why do we need to be moving this quickly and dangerously? Why do we need to spend trillions of dollars overnight? Why do we need to invest half the fucking stock market on this brand new technology as fast as we can? Why can't we develop it in a way that isn't insanely fast and dangerous? Or are we incapable of decisions not based on greed and FOMO?
Who is "we" ? I certainly don't spend trillions on frivolities. I think the Saudis via Softbank do, and these people build fake cities in the desert, they are by definition dumb money.
They earn so much from oil and are so keenly aware this will stop, they'd rather spend a trillion on a failure, than keep that cash rotting away with no future investment.
No project, no country, can swallow the Saudi oil money like Sam Altman can. So, they're building enormous data centers with custom nuclear plants and call that Stargate to syphon that dumb money in. It's the whole business model of Softbank: find a founder whose hubris is as big as Saudi stupidity.
> The other claims that AI will create more jobs than it destroys.
Maybe it's my bubble, but so far I didn't hear someone saying that. What kind of jobs should that be, given that both forms, physical and knowledge work, will be automatable sooner or later?
I haven't seen that either.
That claim just reads like he's concocted two sides for his position to be the middle ground between. I did that essays in high school but I try to be better than that now.
What makes this analogy great is that nobody in the dial up days could imagine Google or YouTube. We’re in the same place now nobody knows who becomes “the Google of AI,” and that uncertainty usually means a new platform is being born.
I would go so far as to say we are still in the computing dial-up era. We're at the tail end, maybe - we don't write machine code any longe, mostly, and we've abstracted up a few levels but we're still writing code. Eventually computing is something that will be everywhere, like air, and natural language interfaces will be nearly exclusively how people interact with computing machines. I don't think the idea of 'writing software' is something that will stick around, I think we're in a very weird and very brief little epoch where that is a thing.
It took a long long time going from a walking bike to the one we know now. It's not going to be different from AI. Transformers will only get us so far and for the rest we need another tock. AGI is not going to happen with this generation of hardware. We are hitting spatial scaling limits in video and image generation and we are hitting limits with LLMs.
Big bias shining through in comparing AI to the internet.
Because we all know how essential the internet is nowadays.
The thing is that the average person now thinks AI is revolutionary. Thus, if you form the analogy correctly, then it tells us that the average person is wrong and that AI is NOT revolutionary. (I'm not arguing either case.)
There is one key way in which I believe the current AI bubble differs from the TMT bubble. As the author points out, much of the TMT bubble money was spent building infrastructure that benefited us many decades later.
But in the case of AI, that argument is much harder to make. The cost of compute hardware is astronomic relative to the pace of improvements. In other words, a million dollars of compute today will be technically obsolete (or surpassed on a performance/watt basis) much faster than the fiber optic cables laid by Global Crossing.
And the AI data centers specialized for Nvidia hardware today may not necessarily work with the Nvidia (or other) hardware five years from now—at least not without major, costly retrofits.
Arguably, any long-term power generation capacity put down for data centers of today would benefit data centers of tomorrow, but I'm not sure much such investment is really being made. There's talk of this and that project, but my hunch and impression is that much of it will end up being small-scale local power generation from gas turbines and the like, which is harmful for the local environment and would be quickly dismantled if the data center builders or operators hit the skids. In other words, if the bubble bursts I can't imagine who would be first in line to buy a half-built AI data center.
This leads me to believe this bubble has generated much less useful value to benefit us in future than the TMT bubble. The inference capacity we build today is too expensive and ages too fast. So the fall will be that much more painful for the hyperscalers.
HN is struggling to understand
Enlighten us.
> Consider the restaurant owner from earlier who uses AI to create custom inventory software that is useful only for them.
That is the real dial-up thinking.
Couldn't AI like be their custom inventory software?
Codex and Claud Code should not even exist.
> Couldn't AI like be their custom inventory software?
Absolutely not. It's inherently a software with a non-zero amount of probability in every operation. You'd have a similar experience asking an intern to remember your inventory.
Like I enjoy Copilot as a research tool right but at the same time, ANYTHING that involves delving into our chat history is often wrong. I own three vehicles, for example, and it cannot for it's very life remember the year, make and model of them. Like they're there, but they're constantly getting switched around in the buffer. And once I started positing questions about friend's vehicles that only got worse.
But you should be able to say "remember this well" and AI would know it needs a reliable database instead of relying on its LLM cache or whatever. Could it not just spin up Postgres in some Codex Cloud like a human developer would? Not today but in a few years?
Why do I need to tell an AI to remember things?! How does AI consistently feel less intelligent than regular old boring software?!
Because you're using it wrong.
Really. Tool use is a big deal for humans, and it's just as big a deal for machines.
Wouldn't an intelligent computer know to use tools? The core of the point being discussed seems to be why do you need to ask it to make it you inventory software when an intelligent system would know that when asked to build an inventory system setting up a database and logging all the information is need and ask agents to do that.
It's the same question that you might have asked in 1920. "This radio hardly works at all. Can't they do something about all the static? I don't see the big deal. This is just a scam to sell batteries and tubes."
A radio isn't intelligent and isn't marketed as such. If you're going to sell software you call intelligent, I don't think I'm out of pocket for saying "this feels even dumber than regular software I use."
I don't find this comparison fitting.
(Shrug) Well, I do. I don't know where you people get the idea that these sorts of things spring forth fully-formed from the brow of Zeus, but it doesn't work that way.
This your first paradigm shift? :-P
A radio is fundamentally do thing the thing it claims. Artificial intelligence is fundamentally not intelligent.
I'm not saying it is useless tech, but no it's not my first paradigm shift, and that's why I can see the difference.
How'd you do at the International Math Olympiad this year?
You seem to bring much more snark than substance.
Sounds like evasion to me. At some point you'll need to explain exactly why systems that thrash Go grandmasters, make entire forums fail the Turing test, and take gold at international math and programming competitions are "fundamentally not intelligent."
The only argument you (and Toucan, who's been around longer) seem to muster is that the systems aren't perfect. They occasionally say stupid things, need extra handholding, babble nonsense and write buggy code, commit blatant plagiarism, fall into fallacious reasoning traps, and can be fooled with simple tricks... unlike people, presumably.
Limited tests in specific settings are always surprising and do speak to it's use, but are not a testament to it being intelligent.
I replied because you answered "you're doing it wrong" to a question of it's failures. It seems you dismiss the concerns of the smaller errors or failures without realizing the point being made. If it's "smart" enough to solve take international math medals and beat grand masters at go, but can't truly understand problems and anticipate needs or issues, to me it's not a genuine intelligence and in it's current form never will be.
It's not that they are not perfect, is that they have no concept of reality, and it's evident in their failures. Beyond this point I am not interested in trying to convince you.
It can do that today. I am doing that today.
"That side of prime rib is totally in the walk-in, just keep looking. Trust me, bro"
More like AI’s Diaper-Up Era aka AI’s Analogy Era to Mask It’s Shortcomings
Great comment threads here, but the OP article leans too much on AI generated text that is heavy on empty synthetic rhetoric and generative AI clichés.
Dial-up suggests he knows that many orders of magnitude of performance increase will happen, like with internet connectivity.
I’m not sure that’s a certainty.
Nice article, but somewhat overstates how bad 1995 was meant to be.
A single image generally took nothing like a minute. Most people had moved to 28.8K modems that would deliver an acceptable large image in 10-20 seconds. Mind you, the full-screen resolution was typically 800x600 and color was an 8-bit palette… so much less data to move.
Moreover, thanks to “progressive jpeg”, you got to see the full picture in blocky form within a second or two.
And of course, with pages was less busy and tracking cookies still a thing of the future, you could get enough of a news site up to start reading in less time that it can take today.
One final irk is that it’s little overdone to claim that “For the first time in history, you can exchange letters with someone across the world in seconds”. Telex had been around for decades, and faxes, taking 10-20 seconds per page were already commonplace.
> Regardless of which specific companies survive, this infrastructure being built now will create the foundation for our AI future - from inference capacity to the power generation needed to support it.
Does that comparison with the fiber infra from the dotcom era really hold up? Even when those companies went broke, the fiber was still perfectly fine a decade later. In contrast, all those datacenters will be useless when the technology has advanced by just a few years.
Nobody is going to be interested in those machines 10 years from now, no matter if the bubble bursts or not. Data centers are like fresh produce. They are only good for a short period of time and useless soon after. They are being constantly replaced.
> from inference capacity to the power generation needed to support it.
Power gen = yes. Reasoning is trivial due to long useful asset life.
For compute, why would it be any different than fiber in the AI bubble pop case? Less useful, but not useless. Are you misremembering the fiber overbuild era? The fiber investments were impaired and sold off for pennies on the dollar. The value here wouldn't go to zero, either.
Really tired of seeing the story about how, “sure Worldcom et al went bankrupt but their investments in fiber optics gave us the physical infrastructure of the Internet today.”
I mean, sort of, but the fiber optics in the ground have been upgraded several by orders of magnitude of its original capacity by replacing the transceivers on either end. And the fiber itself has lasted and will continue to last for decades.
Neither of those properties is true of the current datacenter/GPU boom. The datacenter buildings may last a few decades but the computers and GPUs inside will not and they cannot be easily amplified in their value as the fiber in the ground was.
So weird, I asked AI (Grok) just yesterday how far along we are towards post-scarcity and it replied...
>We’re in the 1950s equivalent of the internet boom — dial-up modems exist, but YouTube doesn’t.
Which is ironic, considering that the 1950s were long before the internet boom. The internet didn't even exist yet, let alone dial-up modems.
I was curious and looked this up: https://en.wikipedia.org/wiki/Modem#1950s
Mass production of telephone line modems in the United States began as part of the SAGE air-defense system in 1958, connecting terminals at various airbases, radar sites, and command-and-control centers to the SAGE director centers scattered around the United States and Canada.
Shortly afterwards in 1959, the technology in the SAGE modems was made available commercially as the Bell 101, which provided 110 bit/s speeds. Bell called this and several other early modems "datasets".
> Benchmark today’s AI boom using five gauges:
> 1. Economic strain (investment as a share of GDP)
> 2. Industry strain (capex to revenue ratios)
> 3. Revenue growth trajectories (doubling time)
> 4. Valuation heat (price-to-earnings multiples)
> 5. Funding quality (the resilience of capital sources)
> His analysis shows that AI remains in a demand-led boom rather than a bubble, but if two of the five gauges head into red, we will be in bubble territory.
This seems like a more quantitative approach than most of "the sky is falling", "bubble time!", "circular money!" etc analyses commonly found on HN and in the news. Are there other worthwhile macro-economic indicators to look at?
It's fascinating how challenging it is meaningfully compare current recent events to prior economic cycles such as the y2k tech bubble. It seems like it should be easy but AFAICT it barely even rhymes.
Yep.
Stockmarket capitalisation as a percentage of GDP AKA the Buffett indicator.
https://www.longtermtrends.net/market-cap-to-gdp-the-buffett...
Good luck, folks.
How valuable is this metric considering that the biggest companies now draw a significant % of revenue from outside the U.S.?
I'm sure there are other factors that make this metric not great for comparisons with other time periods, e.g.:
- rates
- accounting differences
I estimate you’re talking 25% from overseas.
If that bothers you, just multiply valuations by .75
Doesn’t make much difference even without doing the same adjust for previous eras.
Buffett indicator survives this argument. He’s a smart guy.
Besides your chart, another point along these lines is that the article cites Azhar claiming multiples are not in bubble territory while also mentioning Murati getting essentially infinite price multiple. Hmmmm...
> MIT Professor, 1993' quote
words to live by...
Reads like it was written by ChatGPT.
Dial-up was actually useful though.
KIMI just proposed linear attention. I mean, one breakthrough, and blammo, the whole story changes.
The last AI bubble was AI's dial-up era because it was the dial-up era:
https://en.wikipedia.org/wiki/AI_winter
https://www.youtube.com/watch?v=sV7C6Ezl35A
I feel like this article is too cute. The internet, and the state of the art of computing in general has been driven by one thing and one thing alone: Moore’s Law. In that very real sense, it means that the semiconductor and perhaps more generally even just TSMC is responsible for the rise of the internet and the success of it.
We’re at the end of Moore’s Law, it’s pretty reasonable to assume. 3nm M5 chips means there are—what—a few hundred silicon atoms per transistor? We’re an order of magnitude away from .2 nm which is the diameter of a single silicon atom.
My point is, 30 years have passed since dial up. That’s a lot of time to have exponentially increasing returns.
There’s a lot of implicit assumption that “it’s just possible” to have a Moore’s Law for the very concept of intelligence. I think that’s kinda silly.
Moore's law has very little to do with the physical size of a single transistor. It postulates that the speed and capability of computers will double every few years. Miniaturization is one way to get that increase, but there are other ways.
>The internet, and the state of the art of computing in general has been driven by one thing and one thing alone: Moore’s Law.
You're wrong here... the one thing driving the internet and start of the art computing is money. Period. It wouldn't matter if Moore never existed, and his law was never a thing, money would still be driving technology to improve.
> The one thing driving the internet and state of the art computing is money
You're kind of separating yin from yang and pretending that one begot the other. The reason so much money flooded into chip fab was because compute is one of the few technologies (the only technology?) with recursive self improvement properties. Smaller chip fab leads to more compute, which enabled smaller chip fabs though research modeling. Sure: and it's all because humans want to do business faster. But TSMC literally made chips the business and proved out the pure play foundry business model.
> Even if Moore's Law was never a thing
Then arguably in that universe, we would have eventually hit a ceiling, which is precisely the point I'm trying to make against the article: it's a little silly to assume there's an infinite frontier of exponential improvement available just because that was the prior trend.
> Moore's Law has very little to do with the physical size of a single transistor
I mean it has everything to do with the physical size of a single transistor, precisely because of that recursive self improvement phenomenon. In a universe where moore's law doesn't exist, in 2025 we wouldn't be on 3nm production dies, and compute scale would have capped off decades ago. Or perhaps even a lot of other weird physical things would probably be different, like maybe macroscopic quantum phenomena or just an entire universe that is one sentient blob made from the chemical composition of cheeto dust.
Transistor size is not the only metric that matters in computer speed. Maybe you weren't around when 1MHz CPUs were considered fast. Then there were 8Mhz, then 16MHz, then 25MHz, and soon enough it was 250Mhz, then it jumped up to 1GHz, and now we're seeing 4GHz and faster. We're probably not at the end of the GHz that can be achieved. Chip dies got bigger, too. Way bigger. It doesn't matter if a single transistor can't be shrunk smaller than 3nm if the chip size can be increased. We've seen this in Cerebras Wafer Scale Engine (WSE), which is 12 inches by 12 inches and contains 4 trillion transistors. And then there's the possibility of 3D chip design - if you can't go wider, build taller - but the main problem with all of this is heat and power. More transistors, more GHz, larger dies, all means more heat - and heat is the real limiting factor. If heat and power weren't a concern then we'd have far faster computers.
But all of these advancements in processing power are driven by money, not by some made-up "law" that sounds nice on paper but has little to do with the real world. Sorry but "Moore's law" isn't really a "law" in any way like the laws of physics.
You’ve completely ignored my arguments, you’re hung up on one technicality, and now you’re just being derisive. I literally have a degree in computer engineering. I’m well aware there’s more than just semiconductor size. I’m aware of 3D chip fabs. I’m well aware of clock speed as a dimension. I’m also well fucking aware that moore’s law is not a physical law.
My whole fucking point is that neither are the AI scaling laws.
Please stop talking to me.
>The internet, and the state of the art of computing in general has been driven by one thing and one thing alone: Moore’s Law
Your original comment was downvoted quite a bit. Because you're wrong about this statement, and it sticks out more than anything else you wrote.
>Please stop talking to me.
Likewise.
More like Bullshit Era
[dead]
Good point. Experience teaches us that the breakthroughs in AI will lead to something, we just don’t know what that something is yet and there is a lot of (maybe too much) speculative betting on what it could be.
> most people hit a wall figuring out what it's actually good for beyond parlor tricks.
That's what my parents thought about computers and the internet, wondering what it's actually good for beyond burning $9000 in phone bills to Zerg rush Protoss noobs.
And all the other things computers+internet could do, they could already do through other more reliable (at the time) ways.
But then it turned out that simply making mundane tasks just a little bit faster, and reducing the need to interact with strangers by just that little bit, created a new step on the staircase, a new baseline, with which to reach and do other grander things more easily.
Email was the killer app I think that showed everyone how damn useful the internet is. Then Hotmail showed how convenient email can be.
What is the AI version of that? Maybe code generation. Maybe.
Code generation/LLMs are the SMTP and other internet protocols, we don't have killer apps yet.
> What is the AI version of that?
Being able to plan a trip from a single sentence would be one killer app for many people:
"I'm free next week. I'd like to go to A, B, or C for a couple days. What's a cheap flight and a room within this budget near X area?"
and if it could go and also make a booking through your accounts that would be amazing.
But right now even Google's Gemini is an utter useless dumbass if asked to search Google Flights or Airbnb.
I mean if LLMs could just be a natural-language wrapper around existing tools, that'd be amazing in itself. But corporivalry has made that an stillborn dream.
[dead]
[dead]
[dead]
[dead]
[dead]
tl;dr: Gartner Hype Cycle[1]
[1] https://en.wikipedia.org/wiki/Gartner_hype_cycle
[dead]
[flagged]
[flagged]
[flagged]
> I hope you got those chicks on Substack clapping for you, at least. Fast lane to getting laid for sure.
What is this about? Weird thing to say.
I felt the first half of my reply was too elaborate for such a self-promoting, inaccurate, Nostradamus-knows-better kind of post. I thought it was only sensible that I meet nonsense with nonsense in my closing. Looks like you enjoyed that.
It is clearly true. Up until the AI boom, the vast majority of a typical tech company’s costs are software engineers. Now it may be compute costs.
It's not true. Even though the vast majority of a tech company's cost were engineers, they also all grew like crazy and made much more (and grew even more) than they paid in salaries. The OP is correct, if it was just about salaries. These companies would have grown far more and hired far more endlessly, but the complexity of software at that scale makes diminishing returns from hiring more.
You put it best by calling out the scale explicitly.
If anything the push to use AI has made me more expensive as a dev so far. If im honest i am only slightly more productive but i also burn through tokens.
Most of the big services seem to waste so much time clunking through updating and editing files.
I'm no expert but I can't help feeling there's lots of things they could be doing vastly better in this regard - presumably there is lots to do and they will get around to it.
My head canon is that the thing that preemptively pops the bubble is Apple coming out and saying, very publicly, that AI is a dead end, and they are dropping it completely (no more half assed implicit promises).
And not just that, they come out with an iPhone that has _no_ camera as an attempt to really distance themselves from all the negative press tech (software and internet in particular) has at the moment.
Do you know a single person who'd buy an iPhone without a camera? I don't
That's what they used to say about mobile phones with no keyboards :))
Keyboards were replaced with a touch screen alternative that effectively does the same job though. What is the alternative to a camera? Cameras are way too useful on a mobile device for anyone to even consider dropping them IMO.
He's obviously jesting
Oh. Woooosh. Thanks for still being nice about it (-:
AI image generators
Maybe not as an iphone, but they could drop the camera and cellular and make an ipod touch.
That would require people that know about AI to actually choose to cancel it - which nobody that actually knows what AI can do, would ever actually do.
The Apple engineers, with their top level unfettered access to the best Apple AI - they'll convince shareholders to fund it forever, even if normal people never catch on.
Apple at AI is a dead end because Apple sucks at AI, not because its anything about AI