According to @emollick, a new experiment finds that LLM-generated Community Notes are rated as more helpful and less ideological than human-written notes, achieving broader cross-ideological acceptance with higher positive ratings from raters across the political spectrum; as reported by Ethan Mollick on Twitter, the study shows large language model outputs can improve perceived neutrality and usefulness in fact-checking workflows, indicating opportunities to scale moderation quality and reduce partisan rejection rates in social platforms.