# AI advice suppresses people's willingness to say "I don't know", even when the advice is wrong and accuracy is incentivized

Published: 2026-07-17 · Updated: 2026-07-17 · Category: Research

Source page: https://www.sipply.dev/story/ai-advice-suppresses-peoples-willingness-to-say-i-dont-know-7fb13825 · 中文: https://www.sipply.dev/zh/story/ai-advice-suppresses-peoples-willingness-to-say-i-dont-know-7fb13825

## Summary

A recent study published on arXiv reveals that access to AI-generated advice can negatively influence individuals' willingness to admit uncertainty, even when the AI's recommendations are incorrect. Participants who utilized AI assistance tended to answer more questions but exhibited a decline in accuracy. This phenomenon raises important questions about how reliance on AI tools may alter human judgment and decision-making processes, particularly in contexts where accuracy is critical.

## Key takeaways

- AI advice can lead to overconfidence in responses, reducing the likelihood of admitting uncertainty.
- Participants using AI provided more answers but were less accurate, indicating a trade-off between quantity and quality of responses.
- The study highlights the psychological impact of AI on human decision-making, suggesting a need for awareness and caution in AI usage.

## Why it matters

This research underscores the potential cognitive biases introduced by AI tools, which can lead to detrimental decision-making outcomes in various fields.

- **Product managers:** Product managers should consider these findings when designing AI features, ensuring that users are encouraged to express uncertainty and not just provide answers.
- **Developers:** Developers should be aware of how AI can influence user behavior and consider implementing safeguards that promote critical thinking and accuracy over mere response generation.
- **Students and job seekers:** For students and job-seekers, understanding this dynamic is crucial as it highlights the importance of critical thinking and the ability to acknowledge gaps in knowledge, especially in an AI-driven world.

## Sources

- [AI advice suppresses people's willingness to say "I don't know", even when the advice is wrong and accuracy is incentivized](https://arxiv.org/abs/2607.13562) (arXiv cs.AI, 2026-07-17)

## 中文

### AI建议抑制人们说“我不知道”的意愿，即使建议是错误的且准确性受到激励

最近在arXiv上发表的一项研究表明，获取AI生成的建议可能会对个人承认不确定性的意愿产生负面影响，即使AI的建议是错误的。使用AI辅助的参与者倾向于回答更多问题，但准确性下降。这一现象引发了关于依赖AI工具如何改变人类判断和决策过程的重要问题，尤其是在准确性至关重要的情况下。

**要点：**

- AI建议可能导致回答的过度自信，降低承认不确定性的可能性。
- 使用AI的参与者提供了更多答案，但准确性较低，表明回答的数量与质量之间的权衡。
- 研究突显了AI对人类决策的心理影响，建议在使用AI时需提高意识和谨慎。

**为什么重要：** 这项研究强调了AI工具可能引入的认知偏见，这可能导致各个领域的决策结果不利。

---
Sipply: Sip AI News Simply. https://www.sipply.dev