Estonia’s AI Leap programme, a national initiative developed with OpenAI, now reaches roughly 20 percent of the country’s upper secondary students, according to figures reported this week. The programme provides students and teachers with access to AI tools integrated into classroom work rather than treated as an external resource.
Estonia has spent three decades building digital public infrastructure, from electronic identity to online voting, and AI in schools is being approached in the same way: as a system to be deployed nationally, measured and adjusted.
Small country, useful experiment
With a population under 1.4 million, Estonia can run experiments that would take years to coordinate in larger systems. Teacher training, curriculum integration and usage policy can be set centrally and observed across the whole cohort.
That makes the programme interesting beyond its size. Other countries are watching for evidence on questions that remain open: whether AI access improves learning outcomes, how it changes teacher workload, and what happens when students use it for tasks it handles poorly.
The measurement challenge
Access statistics are easy to report. Learning impact is not. Any credible evaluation will need to separate the effect of the tool from the effect of the training and attention that accompanied its introduction, and to track outcomes over more than a single term.
The programme’s design, which emphasizes teacher involvement rather than unsupervised student use, suggests the government is aware of that risk. It also makes the results harder to attribute cleanly.
A template with limits
National programmes depend on conditions that do not generalize: high baseline digital literacy, small administrative scale, and public trust in government technology projects. Estonia’s results will be informative, but copying the model elsewhere will require adapting it to systems with very different constraints.