Terence Tao Shares ChatGPT Conversation on Math Counterexamples

Renowned mathematician Terence Tao publicly shared his actual conversation logs with ChatGPT, demonstrating how he uses large language models to assist in understanding complex mathematical counterexamples. Tech blogger Dan Shipper originally discovered and shared the link to this conversation. The event has garnered widespread attention, as it breaks the stereotype of "don't use ChatGPT to learn math" and showcases the practical potential of AI in serious mathematical reasoning.

Confirmed

According to the records shared by Dan Shipper, while exploring the Jacobian conjecture, Tao tried to get ChatGPT to explain how Fable constructed a seemingly "impossible" or "miraculous" polynomial function. Instead of simply asking the AI for the final answer, he used it as an interactive explanatory tool, continuously making logical connections during the reasoning process and inquiring about the principles behind new theorems.

Why it matters

Several users who reshared the conversation pointed out that this exchange is a "defining moment" in the AI community. User tdhopper believes the true value of this record is that it直观 (intuitively) presents the thinking path of a top mathematician—how to use LLMs to break down logic when facing obscure mathematical counterexamples. User ctjlewis emphasized that this normalizes the practice of top scholars using AI for thought assistance. Furthermore, user Soumitra Shukla pointed out that this demonstrates how the questioning methods of domain experts can massively amplify the model's latent capabilities. Some users even joked whether using Tao's prompts would make ChatGPT think exactly like him.

2026-07-22 ~ 2026-07-23 · 9 related posts

Primary sources