Odobērs: Latvia Needs Systematic Teaching of AI Output Verification

Mārtiņš Odobērs, whose company developed the AI legal platform "Lexu AI," writes that the problem lies not in technology but in understanding how to use it. He notes that AI often gives convincing but false answers, and people trust them because they sound smart. Thus, the key skill is to say: "This answer sounds smart, but I don't believe it."

Estonia launched a nationwide program "AI Leap" in 2025, covering 154 schools, 20,000 students, and 4,900 teachers. The program's goals include not just AI tool usage but also developing critical thinking for democracy, with a budget of 4 million euros. Finland has taught critical thinking about media for over 50 years, and its 2025 plan integrates AI literacy at all education levels.

In recent years, Latvia has taken steps: in April 2026, the Ministry of Education and Science started the first nationwide AI initiative in schools, with 33.4 million euros planned for higher education digitalization by 2029. Riga Stradiņš University developed the first guidelines on AI use in studies, but they are recommendations. The University of Latvia has introduced a controlled AI environment, and from next year, oral exams will be mandatory for final theses. However, no bachelor's program includes a mandatory course on critical evaluation of AI outputs.

Odobērs points out that Latvia already has competence in building responsible AI solutions, such as the "Lexu AI" principle: the system does not provide an answer without a verifiable source. He proposes three steps: expand the school AI program with measurable competence standards, include critical evaluation of AI outputs as mandatory in the digitalization plan, and conduct a national study on students' actual AI use, similar to Estonia's pre-"AI Leap" research. Until the problem is addressed systematically, individual initiatives remain exceptions.