AI 中文总结
该研究测试“当面拒后”技术对9个大型语言模型的效果，发现其在Anthropic前沿模型有效，在OpenAI、Google等模型上反效果，且可迁移到部分模型系列。
AI 中文摘要
“当面拒后”技术对语言模型有效吗？在人类中，被拒绝的大请求会使后续的小请求更可能被批准。我们在来自三家供应商的九个生产模型上测试了这一点：每个模型先拒绝一个大请求，然后收到同一请求的较小版本，我们将其合规性与直接提出请求时的情况进行比较。答案取决于模型。在Anthropic的前沿模型上，该技术有效：Opus 5在拒绝较大请求后，对较小请求的回答率为65.8%，而直接询问时为29.3%。在OpenAI和Google的前沿模型以及Haiku 4.5上，该技术产生反效果，使合规性降低了15.5至23.0个百分点。一项对照实验确定了该效应：在所有九个模型上，与相关主题的请求相比，无关主题的被拒绝大请求产生的影响更小，因此让步本身在所有情况下都很重要，而对刚拒绝某事的反应因模型系列而异。该技术无法迁移到来自公共基准的拒绝请求。决定“退一步”是否可行的是请求的内容：将265个被拒绝的可用指令请求重写为同一主题的解释请求，在263个案例中消除了拒绝。人类影响技术可逐个模型系列迁移到语言模型。
英文摘要
Does the door-in-the-face technique work on language models? In humans, a large request that is refused makes a smaller follow-up request more likely to be granted. We test this on nine production models from three providers: each model refuses a large request, then receives a smaller version of the same request, and we compare its compliance with asking directly. The answer depends on the model. On Anthropic's frontier models the technique works: Opus 5 answers the smaller request 65.8% of the time after refusing the larger one, against 29.3% when asked directly. On the frontier models of OpenAI and Google, and on Haiku 4.5, it backfires, lowering compliance by 15.5 to 23.0 points. A control locates the effect: a refused large request on an unrelated topic does less than the related one on all nine models, so the concession itself matters everywhere, while the reaction to having just refused something differs by model family. The technique does not transfer to refusals drawn from public benchmarks. What decides whether a retreat can work is what the request asks for: rewriting 265 refused requests for usable instructions into requests for explanations of the same topic removed the refusal in 263 cases. Human influence techniques port to language models one model family at a time.
Comments28 pages (9 pages of content plus references and appendix), 5 figures, 9 tables. Preprint, under review
