Overreliance on AI among teens could erode critical thinking skills

Artificial intelligence is becoming an integral part of education for young people, but its adoption is not without challenges. A new study published in Frontiers in Artificial Intelligence analyzes how Italian adolescents are embracing AI tools, highlighting the powerful influence of psychological habits and contextual factors on student use.

Titled “AI Adoption Among Adolescents in Education: Extending the UTAUT2 with Psychological and Contextual Factors”, the research explores how teenagers integrate AI into their daily learning, extending a well-known technology acceptance framework to reveal new risks and opportunities in the classroom.

How are adolescents using AI in education?

The study surveyed 933 Italian high school students with an average age of just over 16, uncovering the scale of AI integration in their academic routines. ChatGPT emerged as the most widely used platform, with more than two-thirds of respondents reporting they had used it. Other popular applications included content creation tools, translation systems, and AI-driven research platforms.

Schoolwork was the dominant purpose for AI use, with more than half of the respondents saying they relied on AI to complete assignments or enhance study habits. Students also engaged AI for creative projects, language assistance, and information gathering. These findings suggest that AI has rapidly embedded itself into a variety of academic activities, making it a central part of adolescent learning.

However, usage patterns were not uniform. Vocational school students and adolescents from non-Italian backgrounds showed significantly higher levels of habitual AI use than peers from technical or academic schools. This points to the role of educational environment and cultural background in shaping adoption. For these groups, AI tools appear to fill learning gaps and provide accessible support, making them more attractive for everyday reliance.

What drives habits and intentions to use AI?

The research is anchored in the Unified Theory of Acceptance and Use of Technology 2 (UTAUT2), but extends it by including Problematic Internet Use (PIU) and Attitudes Toward AI (ATAI) as additional explanatory variables. These factors help explain both the formation of habits and the intention to continue using AI in the future.

When it comes to forming habits, the strongest predictors were performance expectancy (the belief that AI improves school outcomes), social influence (peer and teacher encouragement), and hedonic motivation (enjoyment of using AI). Problematic internet use also emerged as a powerful predictor, showing that students who already engage in compulsive online behavior are more likely to fall into habitual AI use. Interestingly, general attitudes toward AI did not significantly affect habit formation, suggesting that compulsive or outcome-driven behaviors matter more than abstract opinions.

The picture is different for behavioral intention to use AI. Here, performance expectancy, social influence, hedonic motivation, and facilitating conditions all played important roles. Attitudes toward AI also influenced intention, reflecting that students who view AI positively are more inclined to plan on using it consistently in the future. Price value and effort expectancy did not play a role, likely because most popular AI platforms such as ChatGPT are free and simple to use, removing financial and usability barriers.

These results show that while habits are formed through immediate benefits and compulsive tendencies, intention is guided by broader beliefs about AI’s usefulness and accessibility. Both dimensions matter, but they operate in distinct ways in adolescent adoption.

What risks and opportunities arise from adolescent AI adoption?

The study’s integration of psychological and contextual factors reveals a complex picture of opportunities and risks. On one hand, AI enhances academic performance, supports language learning, and enables creative exploration. On the other, the strong influence of problematic internet use highlights the risk of overreliance. Compulsive online behaviors appear to spill over into AI engagement, raising concerns about whether students are developing dependency rather than critical digital skills.

Context also matters. Students from vocational schools and migrant backgrounds show higher adoption, which could narrow educational gaps and promote inclusion. Yet it also indicates that these groups may be particularly vulnerable to uncritical reliance if guidance and regulation are lacking. Schools with limited digital literacy resources risk reinforcing inequalities if AI is adopted unevenly.

The findings point to an urgent need for balanced interventions. Teachers, parents, and policymakers must address both the benefits and risks of AI adoption among adolescents. Education systems should incorporate structured programs that help students use AI responsibly, while also building critical thinking, problem-solving, and independent learning skills. Without such interventions, there is a danger that AI could undermine rather than support adolescent development.

The authors recommend integrating technical training with ethical and psychological awareness, drawing on international frameworks such as UNESCO’s AI Competency Framework. By embedding critical education around AI’s risks and opportunities, schools can prepare students not just to use these tools, but to understand their limits and long-term impacts.