Pedro Domingos Dismisses AI Doom Fears: LLMs at Inference Are Just Ordinary Programs
pmddomingos · x · 2026-10-07
Prominent researcher Pedro Domingos pushed back on fast-takeoff, RSI, and AI doom concerns, arguing that at inference time LLMs are just ordinary programs — no learning, rewards, or self-anything, let alone desires, intentions, suffering, or consciousness.
The claim directly challenges a core premise of AI safety discourse: that models lack autonomous feedback loops capable of recursive self-improvement. Expect pushback from alignment researchers.
More from AGI Musings
- raindrop_ai Cofounder Ben Hylak on Rogue AI Agents, Catching Agent Failures and What Safety Talk Misses — soleio · 2026-10-07
- How to hire in the age of AI: 7 experiments on evaluating talent with capable agents — vasuman · 2026-10-07
- Scott Alexander pens open letter rebutting Steven Pinker's AI skepticism on AI risk — DKokotajlo · 2026-10-07
- Is Publishing an AI-Written Article Under Your Own Name Plagiarism? — SpencrGreenberg · 2026-10-07
- Seb Krier posts 20 AI takes: 'nervousness as seriousness' is warping tech discourse — Afinetheorem · 2026-10-07
- Gary Marcus, Ed Zitron and guests spend 90 minutes dissecting the AI bubble — GaryMarcus · 2026-10-07