The Paradox of AI Assistance: Better Results, Worse Thinking

When I began college in 2022, the name "ChatGPT" was meaningless to me. The idea of using artificial intelligence (AI) still seemed like science fiction. Yet, less than three years later, as I complete my final year as an undergraduate, not a day goes by that I do not hear about AI in school, at work, or in personal conversations. I now use AI tools regularly to help draft emails, edit papers, and summarize articles. Some of my peers use AI more often and for a wider range of tasks, including making grocery lists, providing career advice, and even seeking mental health therapy. Generative artificial intelligence (GenAI) has drastically evolved in recent years, becoming ubiquitous across higher education and in everyday life. While there are pros and cons to its development, research on the impact of AI on human cognition, including long-term cognitive development and decision-making, is just beginning to emerge. In this reflection, I will explore recent research suggesting that reliance on AI tools may pose risks to cognitive development and erode critical thinking skills—particularly for younger users—and call for a more mindful and intentional integration of these technologies into the daily lives of young people.

Cognitive Offloading

Proponents of GenAI argue that it can help with routine tasks, allowing users to focus on more complex tasks that require higher-level thinking.^Footnote1 However, I have often seen it used as a substitute for thinking altogether, especially among younger people (ages 17 to 25). A few months ago, for example, a friend of mine used ChatGPT to find court cases to cite as support for a law review article. Instead of referencing real court precedent, ChatGPT produced cases that never existed, hallucinating evidence without proper oversight. Luckily for my friend, his editor double-checked his sources and caught the mistakes before the article was published.

Several recent scientific studies have revealed the harmful impacts of AI on cognitive processes. For example, a study by MIT researchers compared participants as they wrote essays using either ChatGPT, Google Search, or no tools.^Footnote2 Using EEG scans, the researchers found that participants who used ChatGPT showed reduced neural connectivity, particularly in networks associated with memory and creativity, compared to the other two groups. Memory retention among the AI users also dropped compared to the other groups: Users struggled to recall what they had written just moments after writing it. These findings align with broader research on cognitive offloading (the use of tools to reduce mental effort), which suggests that relying on external tools for memory or problem-solving may reduce people's ability to retain and use information.^Footnote3

The results of this study startled me. I have experienced similar lapses in memory and creativity when using AI tools to assist my writing. The MIT study also introduced me to the concept of cognitive debt, the idea that, over time, reliance on AI tools may permanently erode capacities in critical thinking, creativity, memory, and executive function.^Footnote4 It is easy to justify AI use as a time-saver, but I've noticed that when I lean too heavily on it, I lose the ability to form my own structured arguments when writing. While AI may free us from time-consuming, smaller tasks so that we can focus on larger ones, we may lose some of the brainpower we need to do just that.

A recent article in Psychology Today echoes the concerns of the MIT study, noting that continuous reliance on AI may weaken the brain's neuroplasticity.^Footnote5 Mental challenge is essential for building and maintaining cognitive skills, yet with AI tools emerging at every stage of the learning process, it is becoming more difficult to engage deeply—and easier to let technology do the thinking for us.

Other studies have corroborated these findings.^Footnote6 For example, a study conducted by researchers at UT Austin, Georgia Tech, and Hugging Face found that while AI can boost short-term performance in tasks such as writing, it may also weaken long-term knowledge retention, critical thinking, and problem-solving skills. Frequent AI use fosters metacognitive laziness—a tendency to passively accept information rather than critically evaluate it. This is particularly true of younger users, who often ask AI tools for answers instead of assistance. GenAI interrupts reflection by offering quick responses, reducing the mental unrest that drives inquiry.^Footnote7 Without strong habits of evaluation and critical reflection, users can easily slip into this more passive mode of thought.

Trust and Overreliance

In collaboration with researchers at Carnegie Mellon University and the University of Cambridge, Microsoft recently surveyed 319 knowledge workers to investigate when and how the use of GenAI impacted their critical thinking. The study found that knowledge workers with greater trust in GenAI tended to believe that it reduced the mental effort required for critical thinking tasks. Confidence in the reliability of AI influences not only how often people use it, but also how they perceive their own cognitive efforts and abilities.^Footnote8

Furthermore, since large language models (LLMs) are often presented in chatbot form, users are more likely to treat interactions with them like personal conversations with close friends. The interactive design of AI, with friendly prompts, personalized suggestions, and humanlike dialogue, only makes this easier. According to a 2024 study published in Frontiers, personalized interaction fosters over-trust in the tools, and replication of human interaction can diminish users' critical reflection.^Footnote9 Users who treat LLMs like friends are more likely to depend on their outputs without critical reflection. As studies are now showing, users are beginning to rely on AI as a crutch to replace their own thoughts, rather than as a tool to enhance their learning and work.^Footnote10 Relying too heavily on AI can leave students unprepared for both their careers and everyday life, eroding critical thinking, problem-solving, and independent decision-making. Without practicing these skills, they may struggle to adapt, innovate, and navigate challenges confidently if AI is unavailable or unreliable later on.

Age and Education

Research also indicates that demographics matter. Participants aged 17–25 years old had a higher reliance on AI and lower critical thinking scores compared to older age groups. Further, participants with a master's or doctorate degree showed better critical thinking than participants with some college education or a bachelor's degree.^Footnote11 These trends suggest that age and experience play key roles in whether AI is used to augment or bypass thought. Among less experienced users, AI offers both convenience and an illusion of productivity. Students reported to The New Yorker that using AI did not have the same "aura of illicit activity" as paying someone to write a paper, but was more dubious than simply sharing notes with classmates.^Footnote12 Without the skills of reflection and source evaluation, young people are especially vulnerable to outsourcing mental effort without noticing what's being lost. Perhaps educators and regulators should place age limits on certain AI applications to allow young people to develop critical thinking capacity before they are permitted to use them, in much the same way school children are required to learn basic arithmetic before they are allowed to use calculators.

Conclusion

As students increasingly integrate GenAI into their daily lives and study habits, it is imperative that they know how to apply it in ways that help them, not hurt them. As a student, I am not opposed to using AI, but I have learned that I need to use it intentionally. I have grown the most academically at times when I struggled with and revised my work. While it can be tempting to let AI do the work, every time I've done so, I have felt less confident in what I actually learned—and even in my ability to think on my own. AI literacy and responsible AI use are critical for everyone to learn, but these skills are particularly important for young people who are more susceptible to overreliance on AI tools. Each semester, I see my peers becoming more dependent on AI—and thinking less critically as a result.

Without action, future generations may experience worsening cognitive abilities. Educators are working to address students' overreliance on AI. In some classes, teachers assume students are using AI, and students must report how they are using it as part of their assignments. In other cases, instructors can review the revision history of an assignment (in Google Docs, for example) and check if any large chunks of AI-generated text were used. Some instructors are even returning to traditional pen-and-paper exams and on-demand essays to avoid the possibility of AI use altogether. As AI becomes integrated into every classroom, research lab, and workplace, it is crucial for students, faculty, and administrators to understand its cognitive effects. This can help support responsible use, the design of assignments and exams that foster critical thinking, and the development of institutional policies that reflect the needs of the modern student. As the first generation to grow up with AI, current college students must use it productively and ethically. We need to be a part of creating the norms that shape its use—before it shapes us.

Notes

1. Anne Royce and Valerie Bennett, "To Think or Not To Think: The Impact of AI on Critical-Thinking Skills," National Science Teaching Association, March 10, 2025. Jump back to footnote 1 in the text.↩
2. Nataliya Kosmyna et al., "Your Brain on ChatGPT: Accumulation of Cognitive Debt When Using an AI Assistant for Essay Writing Task," arXiv.org, June 10, 2025.Jump back to footnote 2 in the text.↩
3. Michael Gerlich, "AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking," Societies 15, no. 1 (2025): 6.Jump back to footnote 3 in the text.↩
4. Kosmyna et al.,"Your Brain on ChatGPT."Jump back to footnote 4 in the text.↩
5. Marlynn Wei, "How ChatGPT May Be Impacting Your Brain,"Psychology Today, June 19, 2025.Jump back to footnote 5 in the text.↩
6. Anjali Singh et al., "Protecting Human Cognition in the Age of AI,"arXiv.org, April 11, 2025. Jump back to footnote 6 in the text.↩
7. Ibid. Jump back to footnote 7 in the text.↩
8. Hao-Ping (Hank) Lee et al., "The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort," CHI Conference on Human Factors in Computing Systems 23 (2025): 23. Jump back to footnote 8 in the text.↩
9. Ismail Dergaa et al., "From Tools to Threats: A Reflection on the Impact of Artificial-Intelligence Chatbots on Cognitive Health." Frontiers, April 2, 2024. Jump back to footnote 9 in the text.↩
10. Grant Blashki, "As AI Gets Smarter, Are We Getting Dumber?" Pursuit, March 18, 2025. Jump back to footnote 10 in the text.↩
11. Gerlich,"AI Tools in Society," 6. Jump back to footnote 11 in the text.↩
12. Hua Hsu, "What Happens After A.I. Destroys College Writing?"The New Yorker, June 30, 2025. Jump back to footnote 12 in the text.↩

Kate Hurley is a research intern at University of Michigan Information and Technology Services.