OpenAI staffer says academia unprepared as inference compute decides which fields accelerate
tw_killian · x · 2026-11-10
OpenAI's Will Depue argued academia is unprepared for a world where scientific progress hinges on inference compute: whichever field Stargate's compute points at will accelerate, and talent is already leaching into labs. He proposed a trade where labs recruit genius scientists to design algorithms while they run GPT-6.2 on physics problems via weekend low-priority compute, splitting paper credit for PR value.
The tweet drew 735K views and pushback: UChicago DSI's Niloofar Mire criticized a culture of "disdain for academics" among OpenAI's younger staff, while gabriel1 countered that spotting bullshit is what school is for and that "optimism and crazy ideas" hold 99% of the upside.
More from AGI Musings
- Rumor: Anthropic deliberately capped Fable/Mythos 5.2 research capabilities to avoid triggering RSI — TinfoilTricorn · 2026-11-10
- The AI startup graveyard will be full of impressive but unnecessary products — tech__unicorn · 2026-11-10
- coeff_giving seeks to incubate new progress-oriented orgs funded by AI wealth — Afinetheorem · 2026-11-10
- Miles Brundage shares a video on the challenge of supervising automated AI R&D — Miles_Brundage · 2026-11-10
- Dan Koe's one-person AI content agency: interview first, $2.5K–5K/month first client — xiaohu · 2026-11-10
- One guy, one prompt, $550 — what could a $300M AAA studio build with AI? — TAbrodi · 2026-11-10