Artificial intelligence has become part of everyday student life. From brainstorming essay ideas to explaining difficult concepts, tools like ChatGPT are now helping students complete academic tasks faster than ever before.
That convenience comes with an important question. Are students learning more, or are they simply becoming more dependent on technology?
A growing body of research suggests the answer depends less on the AI tool itself and more on the person using it. New findings show that self-regulation plays a major role in determining whether AI becomes a helpful learning assistant or a shortcut that weakens critical thinking.
The message from researchers is becoming increasingly clear. Students who actively manage their learning process tend to benefit from AI. Those who rely on it too heavily may gain confidence without gaining real understanding.
A new study explored how university students interact with generative AI tools. Researchers from the University of the Basque Country examined responses from 404 students and discovered an interesting pattern.
Students with clear academic goals often showed greater trust in AI. They viewed the technology as a useful way to work faster and achieve objectives more efficiently. At first glance, that sounds like a positive outcome.
The problem emerged when that trust went unchecked. Students who depended heavily on AI were more likely to accept generated information without carefully evaluating it. Instead of questioning responses, they sometimes treated AI output as automatically correct.
Researchers found that strong self-regulation skills helped prevent this problem. Students who demonstrated perseverance, thoughtful decision-making, and a willingness to learn from mistakes were less likely to blindly trust AI-generated answers.
These students continued reviewing information, checking facts, and making independent judgments. Rather than replacing their thinking, AI became one tool among many in the learning process.
Productive Struggle Still Matters
Many students love AI because it provides immediate answers. A difficult question that once required twenty minutes of effort can now receive a response in seconds.
While that speed feels helpful, researchers warn that learning often happens during the struggle itself. Working through confusion, making mistakes, and finding solutions are important parts of skill development.
Evidence from a three-month study involving chess clubs highlights this concern. Participants who had unrestricted access to an AI tutor performed significantly worse than students whose access was carefully controlled.
The difference was dramatic. Students with limited AI access achieved more than twice the performance gains of those who could use the tool whenever they wanted.
Researchers traced the issue to a decline in productive struggle. Students increasingly relied on AI recommendations instead of making decisions themselves. As a result, they missed valuable opportunities to strengthen their own thinking skills.
Interestingly, the problem was not limited to weaker performers. Highly skilled players also requested excessive assistance. This finding challenges the assumption that experience alone protects people from overreliance.
The study revealed that self-regulation matters more than talent. Even capable students can become dependent on AI when they stop monitoring how they use it.
Confidence Without Competence
A study on AI-assisted knowledge work published in April 2026 found that students frequently reported high levels of understanding after using AI tools. Yet objective testing showed that actual learning gains were often much lower.
This gap creates what researchers call a metacognitive disconnect. Students believe they have mastered a topic because an AI explanation made it seem easy. Unfortunately, feeling informed is not the same as being informed.
