13h agoYo Shavit, OpenAI frontier AI safety lead, argues AI models lack the wisdom needed to handle ambiguous tasksHerbie Bradley says RLVR is hard to apply here.SentimentSentimentPos0%Neg100%Users expressed disappointment with current AI tools on ambiguous tasks, noting failures to distill nuanced information or deliver meaningful speed and quality gains in their work.3 comments with sentiment. View comments.