Models Now Outdo a Decade of Human Math — Is This the 'GPT-Style Near-Term AGI'?
teortaxesTex · x · 2026-10-08
Researcher teortaxesTex quotes a stunned "ummm... well.... fuck" post and asks: six years later, do we now have a "GPT-style near-term AGI"? He presses the alignment community: if models are wiping the floor with the last decade-plus of human mathematics, can we still say they are "too dumb to REALLY act on their misalignment" — or is AGI simply being moved out of reach every time capability jumps?
Related event: OpenAI Math Results Spark Polarized Debate Over the Future of the Field(60 posts)→
More from AGI Musings
- AI safety pessimism: “even if we dodge the tech risk, human nature won't save us” — jd_pressman · 2026-10-08
- AI drug discovery hype check: without a simulator, no patent-grabbing windfall — kuza55 · 2026-10-08
- “A recipe is not the chef”: Why AI output isn't expertise — gerardsans · 2026-10-08
- Debate on AI takeover: jagged capabilities, but coherent takeover intent would already show signs, researcher argues — teortaxesTex · 2026-10-08
- AI podcast Roman Forum hits 200K YouTube subs in 8 episodes, Russell and Bostrom lined up — romanyam · 2026-10-08
- P=NP classroom poll flip: from one believer in 2022 to a third of the class in 2025 — aran_nayebi · 2026-10-08