Tao Calls It 'Very Strange and Not Particularly Pleasant'; Gowers Warns of 'Possible Destruction of Mathematical Culture'

The Decoder·high signalXBlueskyLinkedInCopy link

The mathematician reaction to OpenAI's ten results split four ways on August 1: Terence Tao described GPT-5.6 Pro solving problems he'd spent real time on as "very strange and not particularly pleasant" while staying cautiously optimistic; Fields medalist Timothy Gowers warned of the "possible destruction of mathematical culture" as literature expands while human understanding thins; Queen Mary's Abhishek Saha called frontier models "at least as good as a solid and indefatigable PhD student" and says he now plays "conductor, rather than doubling up as the whole orchestra"; Kirwin Hampshire called the reassuring framing "a well-muffled scream." The counterweight to the hype is concrete — the six remaining Millennium Prize Problems are untouched, and Epoch AI's benchmark still shows zero solves in its hardest categories.