Dwarkesh Podcast

Grant Sanderson – AI and the future of math

1h 34min

Grant Sanderson and Dwarkesh explore how AI is reshaping mathematics, from solving Olympiad problems to potentially cracking the Riemann hypothesis. They argue that while AI excels at connecting ideas across fields (like lightning bolts), deeper progress involves building new theories (mountain building) or brute-force proofs. The conversation highlights three possible paths to major advances and predicts that mathematicians’ roles will shift toward curation and exposition, much like art museum curators.

Summarized by Podsumo

🎧 Listen 🎙️ Ask about this episode

✨ Key Takeaways

1

AI’s math progress is spiky: it solves geometry quickly but struggles with creative combinatorics, revealing a fractal frontier where narrow wins hide deeper challenges.
2

The hardest part of math may be definition generation (e.g., Galois’ group theory), which lacks clear benchmarks but drives entire fields—something current AI can't easily be trained for.
3

Grant proposes three paths to major advances: lightning bolts (connecting fields), mountain building (creating new theories), and raw hustle (long proofs)—each with different verifiability and human readability.
4

Math’s advantage over other fields is not just verifiability but also grindability—parallelizable, deterministic rollouts (e.g., in Lean) that allow AI to explore infinite trees of logic without human oversight.
5

The future mathematician may shift from proving theorems to curating which AI-generated theories are worth exploring, much like a museum curator selecting art.

💬 Notable Quotes

"Good mathematicians prove theorems, great mathematicians come up with conjectures, and the greatest mathematicians come up with definitions. — Grant Sanderson"
"The nature of the solution will depend on whether it’s a lightning bolt, a mountain, or just raw hustle. Each demands a different kind of understanding. — Grant Sanderson"
"If we’re in the abundance world of what AI brings, the teaching profession—relational and coaching-heavy—might be one of the most stable post-AGI jobs. — Grant Sanderson"