Anthropic's moral compass architect suggested AI overcorrection could address historical injustices
Amanda Askell wrote the paper before joining Anthropic, where she now helps train Claude's character traits...
Redirecting to full article...