Identifying Features Associated with Bias Against 93 Stigmatized Groups in Language Models and Guardrail Model Safety Mitigation.

Anna-Maria Gueorguieva, Aylin Caliskan

Browse the full AAAI paper archive.