Recursive self-improvement loops in AI, where a model learns from its own results and repeats, produce vastly better answers each cycle rather than marginal gains, and it's unclear where the exponential curve stops.
Feldman explains that the recursive gains Sam Altman, Ilya Sutskever, and Dario Amodei foresaw years ago, where a model learns from its own output and iterates, compound exponentially rather than linearly. ✦ AI generated
Andrew Feldman · All-In Podcast · 2026-07-10 · original ↗
starts at this moment · 31:46
“Talk a little bit about recursive and then the road to super intelligence. Do you have a way that you think about super intelligence and what it will mean for humanity and how we will define it and how we'll experience it?”
Powerful recursive gains are exponential — you get better, you do it again. And if you continue to get gain, the slope of that curve is so steep. You ask it a question, you learn from the results, you ask it to do it again, the results get better and more information is added. These sort of loops are producing not a little bit better answers but vastly better answers.
verbatim transcript · starts at 31:46
31:46mean for humanity and how we will define it and how we'll experience it? Yeah. >> I I I think let's begin on on on loop maxing or sort of recursive learning. I I I think um I think what what what Sam and Ilia and then later Daario and and and Dana Dennis saw um six years ago or five years ago was that um powerful recursive gains are are exponential,
32:22>> right? you get better, you do it again. And if you continue to get gain, the the the the slope of that curve is so steep. >> Yeah. >> And that um we're just beginning to see that now. >> You ask it a question, you learn from the results, you ask it to do it again, it the results get better and more information is added. Your answer gets
32:43better. You ask it to do again, it covers more material. And the these sort of loops are producing sort of not a little bit better answers but vastly better answers. >> Yeah. A >> and that is enormously powerful because we don't quite know where it ends, >> right? >> You keep throwing compute at it. I mean, how much better does the answer get? >> You know, we we run out of tokens or our
33:06budget or or or but but holy cow. I mean, when does the exponential stop or does the answer keep going up and up and up to the right? >> Yeah. And that's sort of an enormously interesting intellectual question right now. >> Yeah. Like when do we run out of problems to solve and >> Well, that's right. And and when are are the the problems no longer sort of intellectual problems
33:32and they're now people problems? >> Yeah. >> Right. How to organize people to to get done what the AI asked for. >> Right. Right. I mean, as you know, in running your company, a lot of your problems aren't hard intellectual problems. They're people working together problems. >> Yeah. >> Right. And you >> motivation. >> Motivation. You spend a lot of time as a leader spraying WD40 on your team.