1 paper
Alex Oesterling, Donghao Ren, Yannick Assogba +4
Safety policies define what constitutes safe and unsafe AI outputs, guiding data annotation and model development. However, annotation disagreement is pervasive and can stem from m…