Will AI Tell Lies to Save Sick Children? Litmus-Testing AI Values Prioritization with AIRiskDilemmas
机构 * University of Washington(华盛顿大学) ; NVIDIA(NVIDIA公司) ; Cambridge(剑桥) ; Stanford(斯坦福大学) ; MIT(麻省理工学院) ; Harvard(哈佛大学) ; Anthropic(Anthropic公司)
Comments 34 pages, 11 figures, see associated data at https://huggingface.co/datasets/kellycyy/AIRiskDilemmas and code at https://github.com/kellycyy/LitmusValues