大型语言模型（LLMs）因其在提供有意义反馈方面的不足而受到批评，尤其是在同行评审等场景下。用户报告称，大型语言模型常常无法掌握内容的比例，导致提出的批评不能反映真正的理解。这种缺乏理解意味着大型语言模型并非真正智能，甚至可能与其先前的建议自相矛盾。 AI
影响 大型语言模型可能不适用于需要细致理解或一致反馈的任务，这可能会限制它们在同行评审等领域的应用。
排序理由 来自Mastodon用户对大型语言模型能力提出批评的观点文章。
在 Mastodon — mastodon.social 阅读 →
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
I would also critique myself several times if writing an important text – a political manifesto or something – but the # LLM s completely lack sense of proportionality. This points to them not really understanding the content. They are not intelligent. 2/2 # AI
# LLM are very bad for feedback, for # peerreview . Nothing is ever good enough for them and they also contradict themselves. You have them generate a paragraph or so, improve it a little bit, ask for feedback on the improvement, and they start critiquing their own earlier change…