AI Summary
5 min readIn mid-May 2024, an AI system solved the Erdős unit distance problem, a decades-old open question about how often a given distance can appear among points in the plane. The result was surprising not just because it was true, but because of how it was done: the model brought in techniques from a 1960s classification, an area unrelated to the problem's usual toolkit. University of Toronto mathematician Daniel Litt calls this his favorite fully autonomous AI result so far, precisely because it felt recognizably creative rather than like a brute-force grind. But as Litt explains in conversation with a16z's Lee Shalev, this kind of success captures only a narrow slice of what mathematicians actually do.
What the Models Can and Cannot Do
Litt is clear: the frontier models (OpenAI's GPT-5.6 Pro and Anthropic's Claude) are now roughly neck-and-neck in mathematical capability. Their outputs look human—recognizable chains of thought, not alien symbol-pushing. They excel at tasks that play to their strengths: long, reliable computations; pulling together technical ideas from many papers a human might not have read; and applying known techniques in clever ways. The unit distance solution fits this pattern—it was a creative application of existing machinery, not the invention of new machinery.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of a16z Show
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:00) **Opening: The Goal of Mathematics** - The opening frames mathematics as the pursuit of understanding, not just producing papers, and introduces the central tension with AI.
- 2 (03:14) **Most Impressive AI Result So Far: The Erdos Unit Distance Problem** - Litt identifies his favorite fully autonomous AI result and explains why it stood out as creative.
- 3 (05:56) **AI's Reasoning is "Very Human," Not Inhuman** - Litt argues that AI mathematical outputs are recognizable and understandable, not alien or superhuman.
- 4 (09:06) **Comparing Claude vs. ChatGPT: Similar Capabilities, Narrow Band** - Litt finds the frontier models are "pretty similar" and solve a "relatively small portion" of what human mathematicians do.
- 5 (13:56) **What Mathematicians Actually Do: Problem Solving vs. Theory Building** - Litt describes the core activities of a mathematician, which go far beyond proving known conjectures.
- 6 (18:39) **How Litt Uses AI in His Daily Work: A Substitute for Google** - For his long-standing research projects, AI is not a deep collaborator but a useful tool for specific, narrow tasks.
- 7 (21:07) **What Drives Mathematicians: Curiosity and "Good Science" Over Beauty** - Litt explains that his motivation is not aesthetic beauty but a desire to understand fundamental concepts.
+ Full timestamped outline available in the app
Show Notes
a16z’s Lisha Li sits down with Daniel Litt, Assistant Professor of Mathematics at the University of Toronto, to unpack AI's rapid progress in mathematics, what today's frontier models can actually do, and what they're still missing about the way mathematicians think.
Daniel explains why some recent AI-generated results are genuinely impressive, including an autonomous solution to the Erdős unit distance problem, but argues that solving problems is only one part of mathematics. Today's models can grind through calculations, combine known techniques, and search enormous spaces, but still struggle with intuition, theory building, identifying the right questions, and developing the kind of big-picture understanding that drives much of mathematical progress.
Lisha and Daniel also explore how AI is already changing mathematical research, why an explosion of AI-generated papers could distort academic incentives, and what happens if researchers outsource the work of thinking rather than use AI to deepen it. Ultimately, they ask a question that extends far beyond mathematics: as AI gets better at intellectual work, how do we make sure humans keep getting better at thinking too?
Resources:
Follow Daniel Litt on X: https://x.com/littmath
Follow Lisha Li on X: https://x.com/lishali88
Stay Updated:
Find a16z on YouTube: YouTube
Find a16z on X
Find a16z on LinkedIn
Listen to the a16z Show on Spotify
Listen to the a16z Show on Apple Podcasts
Follow our host: https://twitter.com/eriktorenberg
Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
More from this podcast
a16z Show →