Voices
Terence Tao proposes labeling problems "analysis-required" so a bare AI answer earns nothing
Tao's Mathstodon warning goes past the speed of AI solutions to the incentives it creates: even a rumor that someone is working on a problem can trigger a large AI push to solve it first, and "the incentives may now be pointing in the direction of no longer sharing any promising research directions with the broader community, which would reverse centuries of traditions of open science." His concrete proposal is an "analysis-required" label so an unexplained AI-produced answer counts for little. Marcus adds that OpenAI spent $22.5M on the Navier-Stokes effort and admitted it "cannot rule out" that de-identified data from customers' use of NYU and Anthropic products helped improve its models.
↳ Follow the thread