Artificial intelligence in education encompasses more than just chatbots, flashy demos, or automated homework help. Its most transformative potential lies in the expensive, repetitive, system-critical work that education systems still largely perform by hand.

This is where the real economics of educational AI begin.

In both Ukraine and Switzerland, large-scale examinations determine access to future educational pathways. In Ukraine, the 2025 National Multi-Subject Test (NMT) involved more than 289,000 participants, of whom 317,091 registered. Testing was organised in Ukraine and abroad, including in 55 cities across 33 countries. In Switzerland, the scale is smaller, but the logic is the same. In the canton of Zurich alone, 4,562 students took the 2024/25 central admission exam for the Langgymnasium, of whom 2,443 passed.

These are not marginal processes. They are recurring, high-stakes, large-scale systems involving selection, certification, evaluation and educational routing.

International evidence already points in this direction. The OECD acknowledges the potential of digital assessment systems to transform assessment practices through technology and smart data, and reports that many countries are already digitising or supporting assessment workflows. It also emphasises that AI and related tools can reduce teachers’ workload by automating lower-value administrative and feedback tasks.

Therefore, the question of whether AI can be used in exams such as the NMT, Matura or Aufnahmeprüfung is no longer relevant.

This is particularly important because high-stakes examinations differ from regular classroom quizzes. While many education systems have digitised standardised evaluations, the OECD notes that high-stakes examinations are digitised less frequently precisely because the demands relating to fairness, reliability and infrastructure are so much higher.

This is precisely why this opportunity is so important.

Ukraine offers a unique combination of scale, urgency and operational complexity. The NMT has already evolved into a large-scale, computer-based examination system distributed internationally under extraordinary conditions. Switzerland offers a different but equally compelling environment, with high institutional trust, strong quality requirements and a culture of procedural fairness. A country like Switzerland is therefore well placed to develop a gold-standard model of trustworthy AI-assisted assessment.

In that sense, the two countries are not opposites. They are complementary testbeds.

Ukraine could become a place where such systems are rapidly scaled up in real-life situations. Meanwhile, Switzerland could become the benchmark for precision, transparency, and ethical implementation.

Together, they could help to define a new European standard for assessment.