Yo Shavit, OpenAI frontier AI safety lead, argues AI models lack the wisdom needed to handle ambiguous tasks · Digg