Cascadity Audio
← AI
0

On Terence Tao's blog, mathematicians debate what their work is for after OpenAI's 700-paper dump

Two guest essays in two days, one practical and one philosophical, from working mathematicians trying to make sense of AI-produced proofs.

Terence Tao's blog ran two guest essays this week on how mathematics should respond to AI. In "Mathematics has changed. Let's act like it" (Nov. 1), mathematician Frank Calegari calls OpenAI's Oct. 6 release of more than 700 research papers a watershed, while saying the papers are "atrociously written," violate the field's norms on attribution, and that some are wrong. He compares the moment to the classification of finite simple groups, which he says cooled interest in that subject, and argues mathematicians should reward people who choose worthwhile problems and understand AI-found proofs, and reform hiring and journals rather than try to change AI companies.

In "Egoless Mathematics in the AI Era" (Nov. 2), Karthik C. S., Amit Sahai and Jie Xue argue that anxiety about being replaced comes from measuring mathematicians by output, and that a truth "has become accessible; its worth is not diminished by the absence of our authorship."

Tao notes both posts were converted to the blog's format using AI.

Read it at the source
Cascadic Analysis 2
UndertowRole underspecification / the Doorman Problem

What hidden risk could pull against this, even if the news is good?

If a mathematician's job were producing proofs, 700 papers at once would look like the job being done. Frank Calegari's complaints point to what that leaves out: he calls the papers "atrociously written," says they break the field's norms on attribution and that some are wrong. He argues the field should reward choosing worthwhile problems and understanding AI-found proofs, which are the parts of the role a paper count never captured.

0
UndertowDelayed Consequence / False Success Problem

What hidden risk could pull against this, even if the news is good?

Calegari says some of the papers are wrong, and finding out which takes careful reading by people, at the slow pace mathematics checks proofs. A flood of results that look finished can be cited and built on before the errors surface. The volume itself makes the review lag longer.

0
Rabbit Holes 1