Over-reliance on chatbots can diminish critical-thinking skills, study finds | US news

A new study from the Massachusetts Institute of Technology is the latest to find that relying too heavily on chatbots can reduce critical thinking skills and potentially reduce our ability to discern misinformation for ourselves.

As AI tools become more sophisticated and accessible, manipulated images and misleading headlines are becoming more common. Artificial intelligence can be part of the solution and has proven useful in helping users detect fake content; But new research suggests that using AI in this way comes at a cost. Over-reliance on artificial intelligence to help understand what is real on the internet can lead to problems in making these decisions.

Released during the four-week study in AprilResearchers followed 67 participants and tested them on whether news-related headline and image pairs were real. They found that AI assistants like Claude and ChatGPT can be useful in detecting fake news; But they became worse at detecting misinformation when participants relied on them too much.

Researchers also found that when it comes to deciding which news headlines and images are real, AI often prioritizes the right response over improving its thinking ability. This addiction can actually worsen judgment in the long run. to work.

“When we interact with AI, we feel like we’re getting better at certain tasks, and there’s enough research to show that we’re not,” says Anku Rani, a doctoral student at MIT and co-author of the study.

Participants in the month-long study were asked to answer questions about fake news and images, with and without the help of an AI assistant running on GPT-4o and integrated with Google search. The chatbot may hint at clues to look for; One example showed the AI ​​chatbot advising the user to take a closer look at a police badge, which revealed that an image was fake.

The study’s authors evaluated how well AI helped participants make the right decisions and how their independent decisions changed over time. They found a compromise: AI helped participants better understand what was real, giving them a 21% higher chance of making the right decision. But when reviewing new images without the AI’s help, their unaided performance worsened by 15.3% in the fourth week of the experiment. “These results suggest that AI may be helpful immediately but may ultimately undermine long-term misinformation detection capabilities,” the study said.

Concerns about the effects of relying too heavily on artificial intelligence, or even other types of technology, are not new. Calculators and GPS devices have dulled the ability to do mental math and navigate neighborhoods without assistance. A. 2025 Lancet study It found that doctors who used AI classification tools to detect cancer got worse over time at doing so on their own. A neuroscientist at the Probability Institute, a metascience research group, recently warned Directing one’s thoughts too much to the AI ​​could weaken the brain’s defenses against dementia.

An analysis of the latest MIT study notes that an AI system’s approach—whether prescriptive or investigative—can impact the user’s ability to make common sense judgments. The study notes that although users often look for a chatbot that will provide speed and precision, it is more nuanced, guided questioning that can improve critical thinking.

The study adds that participants who used AI systems to tell them what to do often said they “complyed with the system because it seemed knowledgeable.” Nearly a quarter of participants said they thought their detection skills had improved even as their performance worsened.

The MIT study has some important limitations. The authors acknowledge that they studied participants predominantly from the US and UK, and that a more diverse sample could indicate whether this skill degradation occurs across cultural contexts and education systems. Longer studies that follow people for more than four weeks could also shed light on whether the effects of over-reliance on AI persist over time.

The researchers say it’s especially important to consider the implications for educators who are increasingly relying on AI for learning tools. Given the vast flow of questionable information online, from news to viral images, from medical claims to political rumors, the study’s observations also apply to the broader public. “As AI becomes increasingly sophisticated, ensuring that these tools foster critical thinking skills rather than cognitive dependency becomes essential to safeguarding the public’s resilience to misinformation,” the study notes.