ResearchtechnologyBrief
If AI can do the math, what do mathematicians do?
What’s it like to watch AI solve a problem you’ve spent years on?

Brief
The a16z crypto podcast episode explores how AI could reshape mathematical research. It opens with researcher Ittai Abraham, who spent about two years on a set of math problems before giving one to an AI model years later. According to the episode, the model made progress within about an hour on a point where Abraham had stalled. Stanford cryptographer Dan Boneh raises a hypothetical in which AI settles P versus NP, with a proof running to a billion pages that a computer could check step by step while people may struggle to grasp it. Abraham, Boneh, Scott Duke Kominers and Justin Thaler discuss proofs that skip key steps, how researchers judge which problems deserve effort, and what to tell a student weighing a five-year math PhD.
Why it matters: The discussion asks what human understanding means when AI can produce proofs that machines check but people may not follow.
Read the full story
Mentioned in this story
Open a coin page for its current price.