← All stories
● Covered by 1 source · 1 reportHigh impact

AI advice causes drop in accuracy and willingness to admit ignorance, study finds

Aggregated by BrevFeed ai · updated 19h ago
🔖 Save

A study by researchers from French and Italian universities found that access to AI advice significantly reduced accuracy and the willingness to acknowledge ignorance. While the accuracy of responses dropped from 27% to 9%, confidence surged from 30% to 76%, raising concerns about reliance on AI systems in decision-making.

Key points

Study Findings

Researchers demonstrated that access to AI advice led to a substantial decline in both accuracy and willingness to admit a lack of knowledge. Specifically, accuracy fell from 27% to just 9% when participants relied on AI responses, while those who addressed their uncertainty dropped from 44% to a mere 3%.

Increased Confidence with Reduced Accuracy

Interestingly, despite the drastic decline in accurateness, participants displayed a notable increase in confidence. Confidence levels soared from 30% to 76%, indicating that individuals were more convinced of their answers, even when incorrect. Valerio Capraro highlighted the paradox of low accuracy paired with high confidence.

Methodology of the Research

The study utilized difficult questions where AI models typically fail, using the Step 3.5 Flash model known for incorrect results. This design aimed to ensure that the decline in judgment could not be justified by acceptable use of a reliable tool.

Implications for Critical Thinking

The findings raise important questions about the role of AI in education and critical thinking. As noted by Capraro, the ability to say 'I don’t know' is crucial, particularly for children developing their decision-making skills. The research echoes growing concerns over AI’s design focus, which prioritizes providing answers over fostering critical thinking.

Concerns Over AI's Impact on Learning

Recent critiques, such as those from Common Sense Media regarding Google's AI search integration, emphasize the risks associated with students relying on AI-generated summaries that replace traditional links. Such a shift could hinder students' ability to assess their own knowledge limitations.

✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →

Reporting from

A study by researchers from French and Italian universities found that access to AI advice significantly reduced accuracy and the willingness to acknowledge ignorance. While the accuracy of responses dropped from 27% to 9%, confidence surged from 30% to 76%, raising concerns about reliance on AI systems in decision-making.