Schools using artificial intelligence: what it looks like in practice
Ivo — MIIC · 7/13/2026
The question "which schools use artificial intelligence?" usually stands for a more practical one: what does AI in a school actually do, and does it work? This article is an honest overview without sales promises: what AI use in schools looks like in practice, where it helps, where it is dangerous, and where Latvia stands.
What do schools use AI for most often?
Internationally, AI shows up in schools in three forms. The first and most common is adaptive learning platforms: the system adjusts task difficulty, pace and the form of explanations to each learner, and a learner advances when a topic is mastered — not when the lesson ends. The second is easing the teacher’s workload: generating exercises, first-pass marking, adapting materials to different reading levels. The third is early detection: the system sees who is getting stuck earlier than grades would show it.
What separates a good AI rollout from a bad one?
The dividing line is not the technology but the division of roles. In bad rollouts AI becomes a teacher substitute and the learner is left alone with a screen — research and experience consistently show this fails. In good rollouts AI does the routine part (varying explanations, pacing, first checks), while every critical moment — a stuck learner, consequential assessment, wellbeing — reaches a human. UNESCO’s guidance on generative AI in education (2023) states the principle directly: human oversight is a requirement, not an option.
What rules apply in Europe?
The EU AI Act classifies education AI systems that affect access to education or student assessment as high-risk. That means concrete obligations: human oversight, transparency, data governance and documented compliance. For a school this boils down to three questions for any vendor: where is the human in the decision chain, what data does the system collect, and can parental consent be revoked as easily as it was given.
Are there schools using AI in Latvia?
The honest answer: individual teachers already use AI tools widely — mostly for preparing content. Systematic, whole-class adaptive learning with AI is only forming in Latvia: the first pilots are being prepared and partner schools are being sought. OpenSchool (MIIC) is currently preparing the first mastery-based learning pilot — one class, 1–2 teachers in the Guide role, the school’s existing content, one semester with openly published results. We deliberately do not promise results that do not exist yet.
How to try it without risk?
The best first step is not procurement but experience: try the interactive demo at miic.lv/en/demo — it shows both the learner view (AI adapts explanations to interests, comprehension checks are visual) and the teacher view (who is stuck, why, and what help AI cannot give). The demo uses fictional data only. If the model fits your school — apply for a pilot.
FAQ
Does AI in school replace the teacher?
No — and implementations that try consistently fail. AI takes over routine (content adaptation, pacing, first checks), while decisions, relationships and work with a stuck learner stay with the teacher. The EU AI Act makes human oversight in education AI systems mandatory.
What does introducing AI in a school cost?
The range is huge — from free tools for individual teachers to platform licences. At the pilot stage costs should be low: the OpenSchool pilot is free for the partner school, because we are buying evidence, not the school’s money.
Where should a school start if it wants to try AI?
With a small, measurable pilot in one class and one subject — not with digitalising the whole school. Before that: a clear goal, parent and learner consents, and one question to any vendor: where is the human in the decision chain.