[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

![NeuroscienceNew Avatar](https://lunarcrush.com/gi/w:24/cr:twitter::32302688.png) Neuroscience News [@NeuroscienceNew](/creator/twitter/NeuroscienceNew) on x 504.5K followers
Created: 2025-07-22 17:28:57 UTC

AI Chatbots Overestimate Themselves, and Don’t Realize It

A new study reveals that AI chatbots are prone to overconfidence, even when their answers are wrong. 

Researchers tested humans and large language models (LLMs) on trivia, predictions, and image recognition tasks, then compared their confidence before and after the tasks. 

Humans adjusted their expectations downward after making mistakes, but AIs often became even more overconfident. 

One model, Gemini, performed worst yet vastly overestimated its own accuracy. 

This inability to self-correct reflects a lack of metacognition in AI systems. 

The findings stress the importance of scrutinizing AI confidence when using these tools.

![](https://pbs.twimg.com/media/GweoSdmWgAATfru.jpg)

XXXXX engagements

![Engagements Line Chart](https://lunarcrush.com/gi/w:600/p:tweet::1947710469738795148/c:line.svg)

**Related Topics**
[trivia](/topic/trivia)
[coins ai](/topic/coins-ai)

[Post Link](https://x.com/NeuroscienceNew/status/1947710469738795148)