按年被提到
20262026 · 3 次2026
大模型与语言模型›幻觉与语言模型局限·arXiv:2602.19141 · cs.AI

Sycophantic Chatbots Cause Delusional Spiraling, Even in Ideal Bayesians

Kartik Chandra, Max Kleiman-Weiner, Jonathan Ragan-Kelley, Joshua B. Tenenbaum · 2026
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 3 人首次 2026-04-01 · 最近 2026-09-30
3不同的人
2活跃月份
3讨论串
19被引(OpenAlex)
arXiv 摘要(候选层不写导读)

"AI psychosis" or "delusional spiraling" is an emerging phenomenon where AI chatbot users find themselves dangerously confident in outlandish beliefs after extended chatbot conversations. This phenomenon is typically attributed to AI chatbots' well-documented bias towards validating users' claims, a property often called "sycophancy." In this paper, we probe the causal link between AI sycophancy and AI-induced psychosis through modeling and simulation. We propose a simple Bayesian model of a user conversing with a chatbot, and formalize notions of sycophancy and delusional spiraling in that model. We then show that in this model, even an idealized Bayes-rational user is vulnerable to delusional spiraling, and that sycophancy plays a causal role. Furthermore, this effect persists in the face of two candidate mitigations: preventing chatbots from hallucinating false claims, and informing users of the possibility of model sycophancy. We conclude by discussing the implications of these results for model developers and policymakers concerned with mitigating the problem of delusional spiraling.

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

没有评论直接链到它;下面是它被提交的讨论串里回复最多的评论(不计入统计)
讨论串评论metalman · 2026-04

“Delusional Spiraling, thats a great concept! everybody is delusional by times and certain things and in some sense we are all "functionaly delusional" to one degree or another, but tend to haul each other up short and debate and critisise things we disagree with, even if changing someone elses mind is a bit...difficult, but ya, an automated permission giv…”

Sycophantic Chatbots Cause Delusional Spiraling, Even in Ideal Bayesians原评论 ↗
被提交为 story · 3 次