EP 264: What AI’s math breakthroughs mean for drug development and genetics

This week on The Genetics Podcast, Patrick goes solo to look at what recent AI breakthroughs in mathematics could mean for biology. He discusses the OpenAI results on the Navier-Stokes equation and hundreds of open math problems, what makes math and code so easy for AI to improve on, how self-driving labs and surrogate endpoints could speed up drug development, and where the biggest gaps and opportunities remain.

Show Notes

0:00 Intro to The Genetics Podcast

00:58 What the OpenAI math breakthroughs could mean for biology

02:57 AI’s recent frontier math results and how mathematicians reacted

09:12 Lean as a proof checker and why it matters for math

10:44 Four conditions behind AI’s math progress and how biology compares

16:22 Why verification speed decides where AI can accelerate biology

19:35 How lab automation shortens feedback loops for variants and antibodies

23:13 How validated biomarkers let early trials predict clinical outcomes

26:11 Why validated surrogate endpoints could speed up drug development

31:58 How startups and big pharma are building self-driving labs

33:40 Virtual cell models built from perturbation data

34:50 Genetics-driven drug discovery and the Phase II success rate question

37:42 Verification speed and surrogate endpoints as biology’s biggest gaps

40:38 Closing remarks