About this episode
<p>Greg Brockman—OpenAI cofounder and Stripe's first engineer—joins John Collison to talk about research-driven product development, an early moment he thought OpenAI was doomed, S curves in AI advancement, and energy bottlenecks.</p><p>Full episode transcript:</p><p><a href="https://cheekypint.transistor.fm/1/transcript">https://cheekypint.transistor.fm/1/transcript</a></p><p><br>Timestamps</p><p>(00:00) Intro</p><p>(02:51) Was OpenAI the first company to take the scaling hypothesis seriously? </p><p>(04:53) Lessons from Dota about deep learning </p><p>(08:08) What is a good new Turing test?</p><p>(08:57) Personalization in AI </p><p>(09:57) Research-driven product development</p><p>(10:26) An early moment OpenAI felt doomed </p><p>(15:01) OS limits on AI product development</p><p>(17:59) When will AI make novel advancements in math or science?</p><p>(20:03) Energy bottlenecks</p><p>(22:30) S curves in AI advancement </p><p>(24:00) AI coding </p><p>(26:25) Refactoring as a killer AI use case</p><p>(27:26) How OpenAI decides what products to built</p><p>(28:53) Growing up in North Dakota</p><p>(30:17) How far away is AGI?</p>