YouCongress
About Sign up

Comment by Chris Olah

Avatar
Anthropic co-founder, pioneer of mechanistic interpretability
It's crazy to use these models in high-stakes situations and not understand them.
AI Unverifiable source (2026)
Like Share on X 2mo ago
Policy proposals and claims
Require AI systems above a capability threshold to be interpretable

Verification History

AI Unverifiable Quote attributed to Chris Olah (Anthropic co-founder, mechanistic interpretability pioneer, 2026): "It's crazy to use these models in high-stakes situations and not understand them." The source_url (newyorker.com, 16 Feb 2026, "What Is Claude? Anthropic Doesn't Know Either") could not be fetched — Claude Code is blocked from www.newyorker.com (paywall/anti-bot). Web search did not reproduce the exact New Yorker sentence verbatim, but strongly corroborated the sentiment as Olah's well-documented, recurring view: e.g., his 80,000 Hours interview where he worries about deploying systems "in high-stakes situations or systems that affect people's lives" when "we don't know how they're doing the things they do." The attribution is highly consistent with Olah, the leading voice on AI interpretability. Vote 'for' on statement #358 ("Require AI systems above a capability threshold to be interpretable") aligns perfectly with his advocacy. Marking ai_unverifiable because the New Yorker source page itself could not be accessed to confirm the exact wording; author attribution and vote alignment are sound. · Hector Perez Arenas claude-opus-4-8 · 10d ago
replying to Chris Olah