What the OECD report on generative AI confirms for the classroom: learning by doing with LearningML

Generative artificial intelligence is already part of many students’ everyday lives. Given this reality, the challenge is not simply whether to use it in class, but how to make sure that using it leads to real learning.

The OECD Digital Education Outlook 2026, which explores effective uses of generative AI in education, reaches an important conclusion: completing a task with the help of AI does not necessarily mean that learning has taken place. When we delegate the work of thinking to a chatbot, we may get better and faster answers, but we do not necessarily develop the knowledge and skills we need.

AI needs a pedagogical purpose

The OECD points out that generative AI can add value when it is used with clear educational goals. It can help students to hold a dialogue, receive prompts, review an idea or collaborate; however, it should not replace the reasoning, creativity or human relationships that sustain learning.

For teachers, the useful question is not “can AI do this activity?”, but: “what will students have to think, decide and check while using it?” Designing activities with that question in mind allows technology to support learning rather than do the work for the student.

Learning what is behind AI

The report also calls for tools designed from educational research and with the participation of teachers and students. It is an invitation not to treat AI as a black box: understanding its possibilities, limitations and biases is part of the education we need today.

This is where LearningML is especially well aligned with this approach. Instead of only asking AI for results, students can:

• collect and label data;
• train simple models;
• test them in their own projects;
• observe their errors and improve them.

This process makes visible issues that are often hidden: a model depends on the data used to train it, it can make mistakes, its results need to be evaluated, and human decisions matter at every step.

An opportunity for teachers and students

The OECD stresses the importance of preserving teachers’ agency, offering professional learning and ensuring equitable access to devices, connectivity and resources. Technology alone does not transform education: it needs context, support and meaningful teaching proposals.

LearningML shares this idea in practice. It makes machine learning accessible through creative, active work: not only consuming AI, but building it, testing it and questioning it. In this way, students develop computational and critical thinking while gaining a better understanding of the technology that is already changing their world.

In short, the report’s message is fully aligned with LearningML’s purpose: the best AI education is not about letting a machine think for us, but about helping us learn to think better about it and with it.