AI writes publishable physics paper in 2 hours, scholars warn credit systems unready
birchlse · x · 2026-10-28
Two physicists asked AI to autonomously pick a problem, solve it, and write a paper worthy of a prestigious journal — it produced a novel, significant result in two hours. Harvard professor Arthur Spirling notes journals, conferences, and tenure processes aren't ready: 'We will pine for the days when we were overwhelmed with obvious slop.'
Related event: Physicists Let AI Pick Its Own Problem, Get Publishable Paper in Two Hours(6 posts)→
More from AGI Musings
- OpenAI proposes all-dimensions Kakeya proof and ETH refutation; expert calls it Nobel-level — ValerioCapraro · 2026-10-28
- AI discourse: coddling humans with sycophantic AI risks creating 'overgrown infants' — DionysianAgent · 2026-10-28
- AI that solves Navier-Stokes still can't write English: the jagged-intelligence debate — AlexTensor · 2026-10-28
- Jack of all trades, master of many: a take on persistent generalist agents — braelyn_ai · 2026-10-28
- Mengdi Wang: AI will become lab infrastructure, turning science into a continuously learning system — MengdiWang10 · 2026-10-28
- "AI Slop" Absolves Humans of Accountability; Call It "AI-Assisted Human Slop" — _vaishnavh · 2026-10-28