AI researchers debate how close we are to recursive self-improvement

New episode with John Schulman, Charlie O’Neill, and Beren Millidge. I got together with some of the most insightful AI researchers I know who are at the openish companies, because I wanted to hear the details of what's actually happening at the frontier and what comes next. 𝐄𝐏𝐈𝐒𝐎𝐃𝐄 𝐋𝐈𝐍𝐊𝐒 * Transcript: https://www.dwarkesh.com/p/john-beren-charlie * Apple Podcasts: https://podcasts.apple.com/us/podcast/ai-researchers-debate-how-close-we-are-to-recursive/id1516093381?i=1000789067132 * Spotify: https://open.spotify.com/episode/0ePd4PUqCpN78hCjVRH0fr?si=wGvk7u5XQwaLfyrvysdIJQ 𝐒𝐏𝐎𝐍𝐒𝐎𝐑𝐒 * Antithesis helps you trust your code. As agents generate more and more of your software, the bottleneck shifts from your engineers actually writing code to verifying it. Antithesis does that testing for you. Ron Minsky, who co-leads Jane Street's tech group, told me that Antithesis was able to help his team shake out bugs in software that had already undergone heavy review. If you want to see how it fits into your development process, go to https://antithesis.com/dwarkesh * Jane Street just launched its most ambitious competition yet: design a protocol-emulator ASIC. Basically, if you have a chip you want to test outside of a live system, you should be able to connect it to your design and have it simulate realistic traffic. Jane Street wants general-purpose, reprogrammable designs that can work across multiple protocols and remain useful as new ones emerge. The most novel submissions will actually get taped out, and the winners will receive a physical copy! The competition is open until January 18, 2027, and teams are encouraged. To get started download the template code at https://janestreet.com/dwarkesh * Grok Bot has been a great way to hand off tasks. My team uses it as a producer: whenever my editor posts a rough cut of an interview in Slack, Grok Bot opens the transcript on its own computer, matches my notes to the exact moments they refer to, and uses a file of my preferences to suggest edits. Then it sends me its top clip candidates so I can review everything from my phone, which saves my editors from sorting through hours of footage. Try Grok Bot for yourself at https://x.ai/bot To sponsor a future episode, visit https://dwarkesh.com/advertise. 𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒 00:00:00 – Steelmanning the case against RSI 00:18:39 – What’s driving the Chinese labs’ progress 00:28:06 – How will automated AI researchers be trained 00:33:51 – Will long-horizon RL elicit AGI? 00:45:24 – The sim-to-real gap 01:00:33 – How much progress is explained by data? 01:18:03 – Why is RL working so well? 01:24:54 – Move 37 and entropy collapse 01:28:32 – Rapid-fire timelines

Dwarkesh PatelhostBeren MillidgeguestJohn Schulmanguest

Sep 11, 20261h 37mWatch on YouTube ↗

EPISODE INFO

Released: September 11, 2026
Duration: 1h 37m
Channel: Dwarkesh Podcast
Watch on YouTube: ▶ Open ↗

EPISODE DESCRIPTION

New episode with John Schulman, Charlie O’Neill, and Beren Millidge. I got together with some of the most insightful AI researchers I know who are at the openish companies, because I wanted to hear the details of what's actually happening at the frontier and what comes next. 𝐄𝐏𝐈𝐒𝐎𝐃𝐄 𝐋𝐈𝐍𝐊𝐒
Transcript: https://www.dwarkesh.com/p/john-beren-charlie
Apple Podcasts: https://podcasts.apple.com/us/podcast/ai-researchers-debate-how-close-we-are-to-recursive/id1516093381?i=1000789067132
Spotify: https://open.spotify.com/episode/0ePd4PUqCpN78hCjVRH0fr?si=wGvk7u5XQwaLfyrvysdIJQ
𝐒𝐏𝐎𝐍𝐒𝐎𝐑𝐒
• Antithesis helps you trust your code. As agents generate more and more of your software, the bottleneck shifts from your engineers actually writing code to verifying it. Antithesis does that testing for you. Ron Minsky, who co-leads Jane Street's tech group, told me that Antithesis was able to help his team shake out bugs in software that had already undergone heavy review. If you want to see how it fits into your development process, go to https://antithesis.com/dwarkesh
• Jane Street just launched its most ambitious competition yet: design a protocol-emulator ASIC. Basically, if you have a chip you want to test outside of a live system, you should be able to connect it to your design and have it simulate realistic traffic. Jane Street wants general-purpose, reprogrammable designs that can work across multiple protocols and remain useful as new ones emerge. The most novel submissions will actually get taped out, and the winners will receive a physical copy! The competition is open until January 18, 2027, and teams are encouraged. To get started download the template code at https://janestreet.com/dwarkesh
• Grok Bot has been a great way to hand off tasks. My team uses it as a producer: whenever my editor posts a rough cut of an interview in Slack, Grok Bot opens the transcript on its own computer, matches my notes to the exact moments they refer to, and uses a file of my preferences to suggest edits. Then it sends me its top clip candidates so I can review everything from my phone, which saves my editors from sorting through hours of footage. Try Grok Bot for yourself at https://x.ai/bot To sponsor a future episode, visit https://dwarkesh.com/advertise. 𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒 00:00:00 – Steelmanning the case against RSI 00:18:39 – What’s driving the Chinese labs’ progress 00:28:06 – How will automated AI researchers be trained 00:33:51 – Will long-horizon RL elicit AGI? 00:45:24 – The sim-to-real gap 01:00:33 – How much progress is explained by data? 01:18:03 – Why is RL working so well? 01:24:54 – Move 37 and entropy collapse 01:28:32 – Rapid-fire timelines

SPEAKERS

Dwarkesh Patel
host
Host of the Dwarkesh Podcast covering AI, economics, and science.
Beren Millidge
guest
AI researcher and CTO of Zyphra.
John Schulman
guest
AI researcher known for work on reinforcement learning and large language model training (formerly at OpenAI).

EPISODE SUMMARY

In this episode of Dwarkesh Podcast, featuring Dwarkesh Patel and Beren Millidge, AI researchers debate how close we are to recursive self-improvement explores recursive self-improvement hinges on objectives, realism, and continual learning The guests steelman why recursive self-improvement might not happen soon: models could keep winning benchmarks while remaining bottlenecked by weak judgment, verification, continual learning, and sim-to-real transfer.

RELATED EPISODES