NormGym Benchmark Shows AI Agents Struggle to Learn Social Norms
Oxford researchers introduced NormGym, a benchmark testing whether AI agents can learn social norms from others' behavior; they found frontier models largely fail to acquire implicit community rules and only learn through punishment feedback.
2026-11-20 ~ 2026-11-20 · 3 related posts
- Study: AI agents barely learn social norms and rely mainly on being punished themselves — ghadfield · 2026-11-20
- NormGym benchmark shows agents learn from own norm violations but not from others — xuanalogue · 2026-11-20
1 near-duplicate retellings: xuanalogue