跳到正文
𝕏 Rohan Paul(x.com)·· 19 天前品質評分44
重點導讀

summary_zh: 哈佛與 MIT 研究指出,語言模型在推理過程中常以「Okay」開頭,導致控制力下降。

  • - 模型:Qwen3-32B、OLMo-3.1-32B。
  • 約 99% 首字機率為「Okay」。

AI 伴隨問答

自備 Key 深度分析

針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:

Rohan Paul
Rohan Paul@rohanpaul_ai
·19 天前
𝕏

New Harvard and MIT paper finds that reasoning models' poor control over their chain-of-thought partly comes from a habit of opening it with "Okay".

Safety teams read a model's reasoning to catch misbehavior, and benchmarks showing models can't steer that reasoning have been reassuring. Why models fail at this was unclear.

The authors told models like Qwen3-32B to reason in uppercase or number each sentence. Qwen3-32B and OLMo-3.1-32B put about 99% of first-token probability on "Okay", so the authors tried banning it or prefilling 1 compliant token.

Banning "Okay" and "Alright" and adding 1 worked example raised average compliance by 34 to 41 percentage points over instructions alone. Tasks that hide information, like avoiding a word, responded less consistently and look like a separate skill across 12 models.

If you trust chain-of-thought monitoring because of controllability scores, separate the opening-word problem from general control and report hiding tasks separately.

推文來源 · X (Twitter)在 X 參與討論

讀者討論

理性交流 · 即時討論