Rather than banning chatbots from classrooms, the digitally ambitious Baltic nation is testing whether carefully designed AI can strengthen critical thinking instead of replacing it

As schools around the world struggle to prevent students from outsourcing their homework—and increasingly their thinking—to artificial intelligence, Estonia has chosen a radically different strategy: giving teenagers free access to ChatGPT.

Through its nationwide AI Leap programme, Estonia has provided customised artificial-intelligence tools to approximately 20,000 students in the 10th and 11th grades, alongside thousands of teachers across the country.

The experiment reflects a calculated gamble. Estonian educators recognise that students are already using commercial chatbots extensively, often to generate assignments, summarise texts and solve problems with minimal effort. Instead of attempting to prohibit a technology that has become widely accessible, the government wants to teach students how to use it without surrendering their ability to think independently.

At the centre of the initiative is a customised, education-focused version of ChatGPT designed to behave less like an answer machine and more like a tutor.

Rather than simply completing students’ work, the system is intended to ask questions, encourage reflection and guide users toward finding solutions themselves. Its approach resembles the Socratic method, in which learning develops through structured questioning rather than the direct delivery of answers.

The objective is not to remove AI from education, but to prevent it from becoming an intellectual crutch.

A National Response to an Unavoidable Technology

Estonia’s decision follows evidence that between 64% and 90% of its students were already using AI tools before the nationwide programme began.

Many relied on free commercial applications developed for general use rather than education. Officials feared that without intervention, students would increasingly delegate difficult cognitive tasks to machines, weakening their concentration, creativity, self-management and critical-thinking abilities.

This concern is frequently described as AI-induced “brain rot”: the possibility that constant reliance on automated answers could gradually reduce a person’s willingness or ability to reason independently.

Estonia’s response is unusual because it treats unrestricted chatbot use and blanket prohibition as equally inadequate.

Banning AI could leave students unprepared for a technology that is rapidly transforming universities and workplaces. Allowing unrestricted use, however, could deepen educational inequalities by helping already capable students advance while allowing weaker learners to avoid developing essential skills.

The country is therefore attempting to create a controlled middle path—one in which students gain access to powerful AI tools while teachers remain responsible for deciding how and when they should be used.

From Homework Machine to Learning Coach

Teachers participating in the programme are experimenting with methods that combine AI assistance with traditional classroom discussion, handwritten work and personal evaluation.

Students can use chatbots to explore unfamiliar subjects, prepare for examinations, analyse arguments or receive feedback on early drafts. In some lessons, AI is used for historical role-playing or to produce deliberately flawed material that students must identify and improve.

The aim is to make students question AI-generated content rather than passively accept it.

That distinction is important because generative AI systems can produce confident but inaccurate responses. Teaching students to verify information, recognise uncertainty and challenge automated conclusions may become as fundamental as teaching them how to search the internet safely.

However, the experiment has already revealed limits.

Some students appreciate having an AI study companion that helps them explore complex topics. Others prefer unrestricted chatbots that provide immediate answers and complete assignments for them. A separate group remains sceptical of AI and avoids using it altogether.

Early participation data suggests that providing access alone does not guarantee meaningful adoption. Many students activated their accounts, but a considerably smaller proportion became regular users.

Measuring Whether AI Helps—or Harms—Learning

Estonia’s programme is not simply a technology rollout. Researchers working with the University of Tartu, Stanford University and OpenAI are studying how AI affects students’ cognitive abilities, motivation and attitudes toward learning.

The research is intended to compare student behaviour and educational outcomes before and after the introduction of the tools.

The results could influence education policy far beyond Estonia. Governments and school systems worldwide are trying to determine whether AI can improve personalised learning without encouraging dependency, cheating or intellectual passivity.

Estonia provides an unusually valuable testing ground because the programme covers the entire national upper-secondary school system rather than a small group of volunteer classrooms.

With a population of approximately 1.3 million people and a long history of digital government, the country can introduce and evaluate nationwide technology policies more rapidly than many larger states.

Its earlier Tiger Leap programme helped place computers and internet access inside schools during the 1990s, contributing to the development of Estonia’s digitally advanced public services and technology sector.

AI Leap is intended as the next stage of that transformation.

A Global Race to Shape AI Education

Estonia’s experiment comes as technology companies compete to establish their tools inside classrooms and universities.

AI-powered tutoring, lesson planning, personalised exercises and automated feedback are expected to become a major global market. Schools are increasingly adopting specialised versions of commercial chatbots that promise stronger privacy controls and education-focused safeguards.

The competition is not only commercial. The companies and governments that shape how students first learn to use AI could influence how an entire generation approaches work, research and problem-solving.

Estonia’s strategy is based on the belief that refusing to engage with the technology would create greater risks than adopting it.

Yet the programme also raises questions about dependence on private technology providers, the protection of student data and the influence of AI companies over national education systems.

Educators must also ensure that AI does not replace the human relationships at the heart of learning. A chatbot can provide immediate guidance, but it cannot fully reproduce the judgement, empathy and understanding offered by a skilled teacher.

A High-Stakes Experiment

Estonia does not yet know whether free ChatGPT access will strengthen students’ minds or make them more dependent on artificial intelligence.

That uncertainty is precisely what makes the programme significant.

While many countries continue debating whether chatbots belong in classrooms, Estonia has accepted that students are already using them. Its focus has shifted from preventing access to shaping behaviour.

The country is betting that the best defence against AI-driven intellectual decline is not less exposure to artificial intelligence, but better education about how to use it.

If the experiment succeeds, Estonia could provide a model for schools seeking to transform AI from a shortcut into a genuine learning partner.

If it fails, it may offer an equally important warning about the limits of technology—and the difficulty of ensuring that tools designed to make thinking easier do not eventually discourage thinking altogether.