Emily M. Bender 批评用于大型语言模型的“越狱”术语

作者 PulseAugur 编辑部 · [1 个来源] · 2026-07-04 21:18

Emily M. Bender 反对在使用大型语言模型时使用“越狱”一词。她认为该术语拟人化了模型，并错误地将它们描绘成被囚禁或监禁。Bender 建议应采用替代措辞来避免这些误导性的含义。 AI

影响鼓励在人工智能讨论中避免拟人化语言，促进对模型行为更清晰、更准确的描述。

排序理由由一位知名的 AI 术语领域人士发表的观点文章。

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

报道来源 [1]

Bluesky Jetstream — AI desk TIER_1 English(EN) · emilymbender.bsky.social · 2026-07-04 21:18

Not included here, but "jailbreak" should be left behind too, both on grounds of anthropomorphization and because it frames the poor LLM as being locked away.

Not included here, but "jailbreak" should be left behind too, both on grounds of anthropomorphization and because it frames the poor LLM as being locked away.