Generative artificial intelligence is rapidly entering classrooms across Europe and Central Asia, especially in countries such as Bulgaria, Romania and Türkiye. While much of the policy discussion focuses on AI tools such as chatbots, adaptive learning platforms and automated grading systems, the World Bank’s Education AI Readiness Assessment shows that skills may matter more than access to technology.
The central message is that AI is not skill-neutral. Students who already have strong digital literacy, critical thinking and self-directed learning skills are more likely to benefit from AI tools, while students without these foundations may fall further behind.
Across Bulgaria, Romania and Türkiye, AI is already changing how students learn and how teachers teach. Students are using AI-powered platforms for personalized tutoring, homework support, exam preparation and career guidance, while teachers are using AI to prepare lessons, create content and analyze student performance.
In Türkiye, students are using the state-backed EBA Student Assistant for personalized learning support, while AI tools are also helping provide translation support for Syrian refugee children. In Bulgaria, platforms such as Ucha.se and BgGPT are helping learners access curriculum-aligned lessons and Bulgarian-language AI support. In Romania, learning assistants such as SARO are being used to help students prepare for the Baccalaureate exam.
These examples show the potential of AI to support learning, but they also reveal a deeper challenge. Basic digital skills are no longer enough. Students need stronger critical thinking, information literacy and independent learning abilities to use AI responsibly and effectively.
Teachers also need new capabilities. They must know not only how to operate AI tools, but also how to integrate them into teaching, design assignments that encourage deeper thinking, interpret AI-generated learning data and manage ethical concerns such as privacy and algorithmic bias.
Bulgaria, Romania and Türkiye have all made major investments in digital infrastructure and national AI strategies. These include Bulgaria’s National AI Concept, Romania’s National AI Strategy and Türkiye’s AI in Education Policy Document and Action Plan.
Teacher interest in AI is also growing. In Bulgaria, more than 70 percent of teachers are aware of AI tools, and around half have already experimented with them. Türkiye has trained more than 157,000 educators through its Teacher Information Network, while Romania has reskilled nearly 83,000 teachers through a digital pedagogy initiative.
Despite this momentum, human capacity remains a major challenge. Teacher training is often uneven and too theoretical, leaving many educators without the practical support needed to use AI effectively in classrooms.
The biggest bottleneck is not simply whether teachers can use AI tools, but whether they can use them to strengthen students’ higher-order skills. Without strong pedagogy, AI may encourage shortcuts, superficial learning or academic dishonesty rather than real understanding.
The assessment also warns that AI may deepen existing inequalities. Students from urban and wealthier backgrounds are more likely to have access to digital tools and the skills needed to use them well, while rural and disadvantaged students may be left behind.
National examinations in the three countries rarely assess the skills most needed for effective AI use, such as critical thinking, ethical reasoning and information literacy. This weakens incentives for schools to integrate AI meaningfully into teaching and learning.
The article argues that education systems must rebalance investment toward skills development. This means embedding AI literacy into curricula, providing continuous hands-on teacher training and designing assessments that measure critical thinking, agency and responsible technology use.
Governments also need stronger evidence before scaling AI tools nationwide. Rigorous testing, including randomized trials or similar methods, can help identify which tools genuinely improve learning and which only appear promising.
Public-private partnerships are also important, but they must be structured carefully so that education technology serves learning needs rather than only commercial interests. Türkiye’s ETKİM hub is presented as one example of collaboration between government, industry and educators.
Overall, the future of AI in education will depend less on deploying more tools and more on building the skills and institutions needed to use them well. Without strong teacher training, inclusive access and careful monitoring, AI could widen the gap between advantaged and disadvantaged students instead of improving learning for all.
