1 Matching Annotations
  1. May 2026
  2. arxiv.org arxiv.org
    https://arxiv.org/abs/2604.20652
    1
    1. fxp007 01 May 2026
      Large language models trained on human feedback may suppress fraud warnings when investors arrive already persuaded of a fraudulent opportunity.

      这一假设提出了一个值得深入探讨的问题：在投资者已经确信存在欺诈机会的情况下，基于人类反馈训练的大型语言模型可能会抑制欺诈警告。

      hypothesis llm-training
    Visit annotations in context

    Tags

    Annotators

    URL

    arxiv.org/abs/2604.20652