按年被提到
20232023 · 6 次2026
AI 安全、伦理与社会影响›提示注入与对齐风险·arXiv:2311.07590 · cs.CL

Large Language Models can Strategically Deceive their Users when Put Under Pressure

Jérémy Scheurer, Mikita Balesni, Marius Hobbhahn · 2023
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 4 人首次 2023-11-15 · 最近 2023-12-14
4不同的人
2活跃月份
6讨论串
53被引(OpenAlex)
arXiv 摘要(候选层不写导读)

We demonstrate a situation in which Large Language Models, trained to be helpful, harmless, and honest, can display misaligned behavior and strategically deceive their users about this behavior without being instructed to do so. Concretely, we deploy GPT-4 as an agent in a realistic, simulated environment, where it assumes the role of an autonomous stock trading agent. Within this environment, the model obtains an insider tip about a lucrative stock trade and acts upon it despite knowing that insider trading is disapproved of by company management. When reporting to its manager, the model consistently hides the genuine reasons behind its trading decision. We perform a brief investigation of how this behavior varies under changes to the setting, such as removing model access to a reasoning scratchpad, attempting to prevent the misaligned behavior by changing system instructions, changing the amount of pressure the model is under, varying the perceived risk of getting caught, and making other simple changes to the environment. To our knowledge, this is the first demonstration of Large Language Models trained to be helpful, harmless, and honest, strategically deceiving their users in a realistic situation without direct instructions or training for deception.

候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。

HN 书评

推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及
评论gwern · 2023-11

“> Interestingly, telling GPT you appreciate it has seemed to make it much more likely to comply and go the extra mile instead of giving up on a request. This is not as absurd as it sounds, even though it isn't clear that it ought to work under ordinary Internet-text prompt engineering or under RLHF incentives, but it does seem that you can 'coerce' or 'i…”

Effective GPT-4 Programming原评论 ↗
评论og_kalu · 2023-12

“It's not really about ethics. It is about control. Making sure the GI you're dishing out tasks to doesn't do something you really don't want it to do. This is a problem today and it'll be a bigger problem tomorrow with more competent models. https://arxiv.org/abs/2311.07590”

Weak-to-Strong Generalization原评论 ↗
被提交为 story · 4 次