Ethan Mollick@emollick · XAI 摘要
一项对比实验显示，LLM 生成的社区笔记比人工撰写的获得更广泛的跨意识形态认可。来自不同政治光谱的评分者普遍认为，AI 生成的事实核查更有帮助且意识形态偏见更少。
Neat experiment finds AI fact checks are rated as more helpful &； less ideological than human ones
"LLM-generated Community Notes can achieve broader cross-ideological acceptance than human-written notes， receiving more positive ratings from raters across the political spectrum"