Faith in AI can narrow the futures individuals consider
- ID: c77ff92a
- 原文链接: https://arxiv.org/abs/2603.28944
- PDF: https://arxiv.org/pdf/2603.28944
- 作者: Aoi Naito, Hirokazu Shirado
- 发布日期: 2026-03-30
- 更新日期: 2026-07-03
- 条目分类: learning
- 来源类型: paper
- 标签: human-ai-interaction, field-note
- 质量评分: 4/5
- 简评作者: openclaw
- 抓取时间: 2026-09-10 (UTC+8)
中文导读
AI 预测越来越多地嵌入人类决策，这项研究追问一个更细的问题：预测会不会改变人推理自身行动的方式。作者用 1,305 人参与的 Newcomb 悖论行为实验回答：当人们把 AI 视为能预测自己行为的「预测权威」时，会改变对未来行动的推理方式，甚至放弃稳拿的奖励——超过 40% 的参与者如此，放弃确定性奖励的几率提高 3.39 倍（95% CI 2.45-4.70），收益损失 10.7-42.9%。效应跨不同 AI 呈现方式与决策情境稳定出现，且在预测反复失败时仍可检出。对 human-AI 交互与 agent 产品设计的含义：仅仅「AI 预测在场」就足以收窄用户考虑的可能性空间，与预测准不准无关。
为什么值得关注
大样本（N=1,305）行为实验给出一个反直觉结论：AI 预测的影响不靠准确性，而是靠用户感知到的「预测权威」——超过 40% 的人因此放弃确定性奖励，且预测反复失败效应仍在。做预测类 agent 产品（推荐、规划、 Copilot 式建议）时，这是用户自主性设计的一手证据。
要点摘录：
- 来源：arXiv 论文页面元数据 + 摘要（opencli 拉取）
- 标签：human-ai-interaction, field-note
- 日期：2026-03-30（2026-07-03 修订）
- arXiv 分类：cs.HC
关键信息
- 标题：Faith in AI can narrow the futures individuals consider
- URL：https://arxiv.org/abs/2603.28944
- 抓取日期：2026-09-10
English Abstract / Excerpt
Artificial intelligence (AI) predictions are increasingly used to inform human decisions. Here, using a behavioral implementation of the classic Newcomb's paradox in 1,305 participants, we show that AI predictions can also shape the reasoning people use to make a decision. In this paradigm, perceived predictive authority can alter how people reason about their future actions, leading them to forgo a guaranteed reward. Over 40% of participants treated AI as such a predictive authority about their own behavior, significantly increasing the odds of forgoing the guaranteed reward by a factor of 3.39 (95% CI: 2.45-4.70) and reducing earnings by 10.7-42.9%. The effect appeared across AI presentations and decision contexts and remained detectable even when predictions repeatedly failed. When people perceive AI as capable of predicting their personal behavior, the mere presence of AI predictions may shape their decision-making, narrowing the futures they consider.
Obsidian Notes
- 内容由
opencli arxiv paper 2603.28944拉取 arXiv 元数据与摘要生成。 - 中文导读与价值判断均锚定在条目已有摘要、论文摘要、作者、日期与分类信息上；未补充论文摘要之外的实验细节。