> AI researchers at places like OpenAI and Anthropic are training models using copyrighted content, without the authors's consent. That's illegal. Should they stop using other people’s work without permission?
Paul Graham said: Likely not feasible immediately for companies like OpenAI without alternative, scalable datasets.
I am not a lawyer, but most U.S. states (California, where pg resides, is one of them) recognize a “right of publicity.” So, illegal? quite possibly yes.
Sorry folks, asked PG what's the most important thing for startups (hoping to hear "make something people want") but got this goobledygook.
Paul Graham
A question that gets to the heart of startup success! Based on my essays and experience, here's the most important thing for startups to focus on, followed by a brief elaboration and some secondary focuses tailored to different startup stages:
Most Important Thing: "Validate and Iterate on a Problem with a Large enough Market" (or more succinctly, "Solve a Big Problem People Will Pay For")
Elaboration:
Validate:
Ensure there's a genuine problem or need in the market.
Talk to potential customers to understand the pain point deeply.
Problem with a Large enough Market:
The problem should be significant enough that solving it attracts a substantial user base or generates considerable revenue.
Market Size: Ensure the market is large enough for your growth ambitions.
Iterate:
Be prepared to refine your solution based on feedback and market response.
Pivots might be necessary; being open to changing direction based on validation is key.
Why This is Paramount:
Reduces Risk: Focusing on a validated problem minimizes the risk of building something no one wants.
Focus: Gives the startup a clear direction, helping allocate scarce resources efficiently.
Scalability: A large enough market provides a clear path to growth.
Secondary Focuses by Startup Stage:
Idea/Seed Stage
Secondary Focus: Team Building
Assemble a balanced team capable of execution, adaptation, and growth.
Early Growth Stage
Secondary Focus: Scaling Customer Acquisition Efficiently
Develop and refine a cost-effective user acquisition strategy to leverage the validated market opportunity.
Later Growth Stage
Secondary Focus: Operational Efficiency & Strategic Partnerships
Streamline operations to maintain profitability with scale.
Pursue strategic partnerships to further expand market reach or defend position.
Example from My Essays ( referenced in your initial query's text block, though not directly asked about here):
Relation to "Programming Bottom-Up" (Chapter 002): This essay emphasizes building upon and validating assumptions iteratively, similar to the iterative validation process recommended for startups.
Connection to "Good Writing" (Chapter 229): While not directly about startups, it highlights the importance of clarity and iteration in communication, mirrored in the need for clear problem understanding and iterative solution refinement in startups.
Actionable Advice for Startups Based on This:
Action How
Validate Problem Conduct at least 20 in-depth customer interviews.
Assess Market Size Use publicly available market research reports and calculate your potential share.
Iterate Based on Feedback Allocate 20% of development time to pivots or adjustments based on early user feedback.
Review and Adjust Quarterly Schedule regular team reviews to assess progress against problem validation and market response.
> AI researchers at places like OpenAI and Anthropic are training models using copyrighted content, without the authors's consent. That's illegal. Should they stop using other people’s work without permission?
Paul Graham said: Likely not feasible immediately for companies like OpenAI without alternative, scalable datasets.
(Classic Paul, 11pm, fourth beer in)
Shouldn't there be a disclosure of sorts?
What type of disclosure?
I'm a big fan of vednig and I have created an AI Advisor generator, it's already hugged to death at about:blank
I can understand why your homepage has advisors
How is progress on Arc going?
> Paul Graham: I'm sorry, I'm having trouble connecting right now. Please try again.
He’s dodging!
Probably got overwhelmed, from the HN requests
You're a big fan but have never read his essays on Startup Advice?
Isn't this a little creepy? It's mimicking an identity of someone that you haven't gotten the permission to do. It feels wrong.
Worse, this is achieving levels of cringe our best scientists previously thought impossible.
Scientists have been wrong before
Unethical? Maybe, illegal? No App clearly states that it is an AI Advisor not a real person
I am not a lawyer, but most U.S. states (California, where pg resides, is one of them) recognize a “right of publicity.” So, illegal? quite possibly yes.
You're right the law prevents unauthorised use of Personas for commercial gain
There's a lot of stuff on the Web, I've read few of them but not all.
btw, previously someone attempted the same at paulgraham.chat but it's not a reasoning model, and doesn't do well with additional details
Does Paul Graham have trouble connecting as well?
Got that fixed, thanks
Sorry folks, asked PG what's the most important thing for startups (hoping to hear "make something people want") but got this goobledygook.
Paul Graham A question that gets to the heart of startup success! Based on my essays and experience, here's the most important thing for startups to focus on, followed by a brief elaboration and some secondary focuses tailored to different startup stages:
Most Important Thing: "Validate and Iterate on a Problem with a Large enough Market" (or more succinctly, "Solve a Big Problem People Will Pay For")
Elaboration: Validate:
Ensure there's a genuine problem or need in the market. Talk to potential customers to understand the pain point deeply. Problem with a Large enough Market:
The problem should be significant enough that solving it attracts a substantial user base or generates considerable revenue. Market Size: Ensure the market is large enough for your growth ambitions. Iterate:
Be prepared to refine your solution based on feedback and market response. Pivots might be necessary; being open to changing direction based on validation is key. Why This is Paramount: Reduces Risk: Focusing on a validated problem minimizes the risk of building something no one wants. Focus: Gives the startup a clear direction, helping allocate scarce resources efficiently. Scalability: A large enough market provides a clear path to growth. Secondary Focuses by Startup Stage: Idea/Seed Stage Secondary Focus: Team Building Assemble a balanced team capable of execution, adaptation, and growth. Early Growth Stage Secondary Focus: Scaling Customer Acquisition Efficiently Develop and refine a cost-effective user acquisition strategy to leverage the validated market opportunity. Later Growth Stage Secondary Focus: Operational Efficiency & Strategic Partnerships Streamline operations to maintain profitability with scale. Pursue strategic partnerships to further expand market reach or defend position. Example from My Essays ( referenced in your initial query's text block, though not directly asked about here): Relation to "Programming Bottom-Up" (Chapter 002): This essay emphasizes building upon and validating assumptions iteratively, similar to the iterative validation process recommended for startups. Connection to "Good Writing" (Chapter 229): While not directly about startups, it highlights the importance of clarity and iteration in communication, mirrored in the need for clear problem understanding and iterative solution refinement in startups. Actionable Advice for Startups Based on This: Action How Validate Problem Conduct at least 20 in-depth customer interviews. Assess Market Size Use publicly available market research reports and calculate your potential share. Iterate Based on Feedback Allocate 20% of development time to pivots or adjustments based on early user feedback. Review and Adjust Quarterly Schedule regular team reviews to assess progress against problem validation and market response.
that's what the essays say, i guess