Greg Burnham

What Greg Burnham said

1. The TWIML AI Podcast · Sep 29, 2026

   AI Progress Explained: From Math Olympiads to Navier-Stokes

   8 moments · 47 min

   6:49
   1. 20:16

      AI math has advanced from school problems to open research ProPro pick

      “AI can solve problems that humans have cared about just independently.”

   2. 30:45

      AI uses accumulated knowledge and persistence in hard math

      “One is it knows everything. It knows all the prior work. It's got the literature memorized.”

   3. 7:44

      Benchmark gains mix targeted training and generalization ProPro pick

      “that deep generalization means you could really see, if that happened in a rich way where you just didn't have to even be trained on something at all to gain capability in that area”

   4. 14:56

      Compute and data underpin most measured AI progress

      “until you see the data and compute supply chain ramp significantly or discontinuously, to be more precise, you're not likely to see discontinuities on the model side.”

   5. 36:12

      Experts judge novelty by how much of a hint humans needed

      “How big a hint would a human have needed to solve this in hindsight?”

   6. 54:13

      Current harnesses do not make models learn strategy on their own

      “we've tried pretty hard and have mostly not found anything that cracks this”

   7. 49:40

      Models still struggle with open-ended work and learning by practice

      “gpt 5.6 scores x out of the box and then stays flat on its playthroughs”

   8. 44:12

      Automatic checks make unsolved math problems usable as benchmarks

      “This is maybe the big piece of work on our side, was to make it easy to verify automatically whether an AI system has solved this problem.”