Pedro Domingos: guardrails make AI stupid
pmddomingos · x · 2026-11-01
Pedro Domingos, author of "The Master Algorithm," posted a terse critique: "Guardrails make AI stupid," arguing safety guardrails come at the cost of model capability.
It's a concise statement of his long-standing position in the ongoing safety-vs-capability debate in AI.
More from AGI Musings
- Instinct says businesses are opening private APIs and whitelisting its IPs to capture agent traffic — mon__lim · 2026-11-01
- Pedro Domingos: Recursive self-improvement already exists — it's called machine learning — pmddomingos · 2026-11-01
- Box CEO Aaron Levie: AI output should now vastly exceed what you could do alone — 2C_ornot2C · 2026-11-01
- Pedro Domingos mocks Anthropic's security plan: build dangerous AI, then block white hats — pmddomingos · 2026-11-01
- AI to cure aging and depression? Optimist thread meets skepticism on drug pipelines — rand_longevity · 2026-11-01
- AI doomers are wrong: legacy software modernization will create huge demand for builders — enggirlfriend · 2026-11-01