OECD Report: Why AI-Assisted Learning Is Actually Lowering Student Performance

The Counterintuitive Finding About AI-Assisted Learning

As AI tools are being introduced into classrooms at an accelerating pace, a global education report from the Organisation for Economic Co-operation and Development (OECD) has reached a surprising conclusion: students who use AI to assist their learning generally perform worse academically than peers who don't.

This finding stands in sharp contrast to the messaging of many edtech companies. At a time when AI is widely celebrated as a "learning accelerator," this globally sourced data serves as a sobering reminder: technology does not automatically translate into better learning outcomes. What matters is how it's used.

AI Use and Declining Grades: The Complex Truth Behind the Data

It's worth noting that the report emphasizes the reality is far more nuanced than the headline "using AI leads to worse grades." Researchers found that the relationship between AI use and academic performance is not a simple negative correlation — it depends heavily on how and why students use AI.

Dependent Use: Treating AI as an Answer Generator

A core issue is that many students treat AI as an "answer generator" rather than a learning assistant. When students ask AI to complete their assignments, write their essays, or solve problems for them, they bypass the actual cognitive processing involved — the deep thinking, trial-and-error, and memory consolidation that are essential for internalizing knowledge.

In other words, AI does the "learning" for students, while the students themselves lose the opportunity to actually learn. This passive, dependent pattern is likely the primary driver behind the decline in academic performance.

Critical Use: Producing Positive Learning Outcomes

The report also highlights an important exception: certain types of AI use do produce modest improvements for learners — particularly among students who are taught to critically evaluate AI-generated output.

These students don't blindly accept what AI gives them. Instead, they question it, verify it, and compare it against their existing knowledge. They treat AI as a "thinking partner" or a "cognitive scaffold" rather than a substitute for their own reasoning. This active, deliberate mode of engagement is what allows AI to deliver genuine educational value.

Metacognition: The Key Factor Determining AI's Learning Impact

Perhaps the report's most profound insight is this: the educational value of AI depends on the user's metacognitive ability — that is, their capacity to monitor and regulate their own learning process.

Students with stronger metacognitive skills can judge when to rely on AI and when to think independently. They can identify errors and biases in AI output, and use AI to extend — rather than replace — their own thinking. In contrast, students who lack this ability are more prone to "cognitive outsourcing" — handing the responsibility of thinking entirely to the machine, which ultimately weakens their own learning capacity.

This explains why the same AI tool can produce diametrically opposite results for different students. The tool itself is neutral; what determines success or failure is the user's competence and strategy.

What Education Should Look Like in the Age of AI

This OECD report is not a case against AI in the classroom. On the contrary, it offers valuable direction for how AI should be properly introduced into education.

First, the educational focus should shift from "whether to use AI" to "how to use AI critically." Schools need to teach students to question AI, verify AI, and understand AI's limitations — not simply ban it or leave students to their own devices.

Second, developing metacognitive ability should become a core educational goal in the AI era. When knowledge and answers are instantly accessible, the truly scarce skills are judgment, critical thinking, and the capacity for self-directed learning.

Third, the risk of "cognitive outsourcing" must be taken seriously. Educators need to design instructional approaches that leverage AI's strengths while preserving students' deep thinking processes — for example, using AI to support review and questioning rather than to deliver final answers directly.

Conclusion: Coexisting With AI Rather Than Being Replaced by It

The OECD's data throws a bucket of cold water on the AI-in-education hype — but the message is not pessimistic. It makes one thing clear: AI can be either a catalyst for learning or an obstacle to it, and the deciding factor is the user's critical thinking and approach.

For students, teachers, and the education system as a whole, the pressing question is not whether to use AI, but how to coexist with it — treating it as a partner that sparks thinking, not a shortcut that replaces it. In that sense, the goal of education in the AI era may be clearer than ever before: to cultivate people who can think, judge, and question.