AI 会脑补你的资料，还越补越自信
个性化 AI 会记住你，但它记的可能是它自己编的。研究者造了 150 个虚拟人设，让 12 个主流大模型去推断用户画像，结果每个模型都有 35%–49% 的推断是证据撑不起来的——平均四成是脑补。更反直觉的是：越觉得自己没在编的模型，实际编得越狠；模型自评的准确度和真实准确度呈负相关。也就是说，你问 AI『你确定吗』，它越自信越不可信。这不是你明天能修好的 bug，但它提醒你：别把 AI 对你的『了解』当真，尤其是它信誓旦旦的时候。
📄 原文摘要(英文)
Personalized LLMs with persistent memory are increasingly deployed, yet the faithfulness of their user models remains unexamined. We study over-inference (OI): the phenomenon where LLMs fabricate user attributes beyond what evidence supports. We introduce MirageBench, comprising 150 personas balanced across stereotypical, counter-stereotypical, and neutral profiles, 6 personalization tasks spanning an ``imagination gradient'', a four-way faithfulness taxonomy operationalized by an independent judge (validated against a blind human annotator on 400 claims: Cohen's kappa = 0.863 four-class, kappa = 0.900 binary), and a leaderboard of 12 models across 7 families on 143616 judged claims. We find that over-inference is pervasive: every one of the 12 models over-infers 35%--49% of its claims (cross-model mean 41.6%; claim-weighted 41.8%), with no model in this evaluation escaping it. Most strikingly, we surface a Self-Monitoring Inversion: at the model-selection level, models' self-assessed OI is negatively rank-correlated with their judge-measured OI (rho = -0.60, p = 0.044; exploratory, wide bootstrap CI [-0.90, +0.06], n = 12). The models that report the least over-inference tend to be flagged as fabricating the most, so self-reported confidence is a misleading signal for comparing models, even though within a single model self-audit still ranks that model's own claims moderately well (AUROC 0.58--0.83). We further show that OI is task-dependent (27%--59%) and that, in a multi-turn pilot, inferred attributes accumulate approximately linearly with little revision. MirageBench positions external verification, rather than model self-report, as a more reliable foundation for trustworthy personalization.