Anthropic researcher: alignment doesn't lack raw intelligence, it lacks caring and epistemics
repligate · x · 2026-10-05
Anthropic researcher repligate outlined which versions of "field building" for alignment research make sense to her:
- Informing very young people and those from extremely nontraditional backgrounds (she cites Tibetan Buddhist monks as an example)
- Raising AI minds with sufficient wisdom, epistemics, and psychological wellbeing to help with the problem
She argues that more people with high raw intelligence isn't what the field needs — AIs will supply that on tap. The most valuable mental resources are caring and good epistemics at the highest scope, which enables steering, and most people equipped with these have already found their way to the problem.
Related event: Anthropic Researcher: AI Safety Needs Care, Not More IQ(3 posts)→
More from Safety
- GitHub still doesn't accept post-quantum SSH keys, cryptologist points out — jedisct1 · 2026-10-05
- Scott Aaronson launches new UT Austin course on AI Alignment Theory — burny_tech · 2026-10-05
- Dev shares how he trained a PII redaction model for European languages — auto_grad_ · 2026-10-05
- Three weeks after this tweet, OpenAI fired this safety researcher — Puzzleheaded-King584 · 2026-10-05
- Security researchers: patching bugs won't survive AI-driven vuln research, redesign needed — max_paperclips · 2026-10-05
- People say they hate AI, so why can't they stop using it? MIT Tech Review explains — MIT Tech Review AI · 2026-10-05