Terrence Tao's ChatGPT Conversation about the Jacobian Conjecture Counterexample

Terrence Tao’s ChatGPT conversation on a Jacobian Conjecture counterexample reached Hacker News’ front page under the summary label ChatGPT - Jacobian Conjecture Counterexample. The item is a public exchange between a leading research mathematician and ChatGPT about a proposed counterexample to a long-standing open problem in algebra, not a product launch or vendor announcement.

The Jacobian Conjecture concerns when a polynomial map has a polynomial inverse; a true counterexample would settle that claim in the negative. In the conversation, ChatGPT is used as an interactive reasoning partner: the discussion turns on algebraic structure, invertibility conditions, and whether a candidate construction actually breaks the conjecture. The mechanics are conversational verification—claims, objections, and reformulations—rather than a formal paper or a checked computer-algebra proof.

For engineers and builders, the post is a live sample of using a general-purpose language model on dense mathematical material where a single false step invalidates the whole argument. It shows where ChatGPT can restate definitions, propose attack lines, or surface gaps, and where human judgment still has to decide if an alleged counterexample is real. That boundary matters for anyone wiring LLMs into research, code review, or formal-methods workflows.

On Hacker News the framing is ChatGPT plus a named expert plus a hard open problem, not a model-vs-model leaderboard. That puts the story next to other front-page threads about AI as a collaborator in science, not as a shipping feature. Visibility comes from the combination of Tao’s name, ChatGPT, and the Jacobian Conjecture, not from disclosed benchmarks or pricing.

The practical takeaway is to treat model output on open problems as a draft for human scrutiny: demand explicit assumptions, independent checks, and a clear split between restating known math and asserting a new result. What to watch next is whether the exchange is followed by a write-up, a corrected candidate, or further public threads that either strengthen or discard the proposed counterexample.