In a world where artificial intelligence is increasingly integrated into our daily lives, a new study sheds light on its impact on human cognition. Researchers from French and Italian universities have discovered that the availability of AI advice can suppress critical thinking, making individuals less likely to admit when they do not know something.
Valerio Capraro, an associate professor at the University of Milano-Bicocca, emphasized the importance of acknowledging the limits of one’s knowledge. He noted, “For humans, the capacity to say, ‘I don’t know,’ is very important because it represents the recognition of the limits of our own knowledge.” The researchers sought to understand whether easy access to AI-generated answers would interfere with this capacity.
Their findings are encapsulated in the title of their paper: “AI advice suppresses people’s willingness to say ‘I don’t know’, even when the advice is wrong and accuracy is incentivized.” To investigate this, Capraro and his colleagues designed a series of questions that large language models typically struggle to answer accurately. Participants were asked about specific visual details in films, such as the color of uniforms in Bend It Like Beckham or the vehicle driven by Monica in Like a Cat on a Highway.
Using the Step 3.5 Flash model, which is known for its inaccuracies, the researchers divided participants into two groups: one that answered without AI assistance and another that could consult the AI. The results were striking. In the baseline scenario, 44 percent of participants admitted they did not know the answer. However, when AI advice was available, this figure plummeted to just three percent.
Moreover, the accuracy of responses also declined. While 27 percent of participants answered correctly without AI, only nine percent did so when they relied on AI assistance. This led to a paradoxical increase in confidence; participants’ belief in their correctness soared from 30 percent in the baseline to 76 percent with AI help, despite the potential for AI-generated inaccuracies.
The researchers also explored the effects of monetary incentives, which slightly improved the willingness to admit ignorance and the accuracy of responses. However, even with these incentives, the numbers remained significantly lower than the baseline.
While the study focused on film trivia, the researchers believe their findings have broader implications across various domains. Capraro expressed concern about the societal ramifications, particularly for children who grow up with AI systems. He advocates for enhanced AI literacy and educational initiatives to ensure that future generations develop essential critical thinking skills.
This article was produced by NeonPulse.today using human and AI-assisted editorial processes, based on publicly available information. Content may be edited for clarity and style.
