Doomerism is a category error: aligning arbitrary AI vs specific AI
inductionheads · x · 2026-09-11
A researcher argues AI doomerism confuses the difficulty of a problem class with its instances: proving alignment for arbitrary future AIs failed, but that says nothing about aligning a specific one — akin to declaring all NP problems unsolvable after finding one NP-complete problem.
More from AGI Musings
- Jensen Huang slams ex-Anthropic researcher's warnings as 'outlandish' as generational AI doomerism debate heats up — nptacek · 2026-09-11
- Gary Marcus: 'rationalist' now means believers in an AI ghost story — GaryMarcus · 2026-09-11
- VC challenges OpenAI researcher to bet on humans not going extinct by 2030 — joelbot3000 · 2026-09-11
- Anthropic capabilities employee: AI poses moderate extinction risk — nabla_theta · 2026-09-11
- Anthropic capabilities team member says he joined to reduce AI extinction risk — EigenGender · 2026-09-11
- 'AI Articulates My Views Better Than I Can': Reddit User on Talking with AI — grateful2you · 2026-09-11