For its 25th edition, European Schoolnet's annual conference, Eminent, took place in Madrid on 11–12 December 2025 with a deceptively simple question at its core: when does artificial intelligence make sense in schools?

At a time when AI tools are spreading faster than policy frameworks, classroom guidance or research evidence can keep pace, the question was deliberately framed not around innovation for its own sake, but around purpose, responsibility and impact. Organised by European Schoolnet in collaboration with the Ministry of Education, Vocational Training and Sports of Spain, and supported by EU-funded projects Agile EDU and EmpowerED, Eminent 2025 brought together policymakers, researchers, practitioners and system leaders from across Europe to take stock  and to look ahead.

Beyond hype: education, humanity and purpose

Opening the conference, Marc Durando, Executive Director of European Schoolnet, invited participants to pause at a critical juncture. After several years of rapid experimentation across Europe, he argued that the time had come to step back and examine what has already been put in place and what still needs careful attention. "Technology alone does not transform education," he reminded the audience. "What matters is identifying where it truly makes a difference, and under which conditions."

These reflections set the scene for the first keynote, delivered by Olli-Pekka Heinonen, Director General of the International Baccalaureate Organisation. Rather than framing AI as either a threat or a solution, Heinonen challenged participants to move beyond binary thinking. "AI is neither good nor bad by default," he observed. "It is both and it is our responsibility to decide when it serves the purpose of education."

Drawing on global perspectives and recent research, Heinonen argued that the most profound impacts of AI in education may be indirect. As AI systems increasingly outperform humans in certain cognitive tasks, education systems are confronted with deeper questions: what does learning really mean, what should be assessed, and which human capacities should be strengthened?

He pointed in particular to creativity, emotional development, social interaction and embodiment as areas where human learning remains fundamentally distinct. Learning, he reminded participants, does not happen in isolation or purely in the mind: "It happens in the interaction between the learner and the world." In this light, AI should not replace thinking but should be used selectively to support human development and flourishing.

AI, data and the reality of student use

the first roundtable discussion, which focused on the responsible use of AI for the benefit of students, brought together evidence from research, policy and practice.

Drawing on emerging international evidence, Quentin Vidal, Policy Analyst at OECD highlighted a crucial distinction: improved performance does not automatically translate into improved learning. Studies show that while students may produce more polished work with generative AI, learning gains can diminish once the tool is removed. "We need to distinguish carefully between doing better on a task and actually learning better," he cautioned.

At the same time, research also points to promising uses of AI as a collaborative or creative partner; particularly when tools are designed with pedagogical intent and guided by teachers' professional judgement. Øystein Gilje, Professor at the University of Oslo and Katarina Sperling, Researcher at the Artificial Intelligence, Simulation and Teaching Laboratory at Linköping University shared insights into how young learners are already using AI extensively, often outside school and at increasingly early ages. Their findings illustrated both creativity and risk, from homework support to emotional reliance, underscoring the urgency of guidance and critical literacy.

Within this discussion, Lidija Kralj, Senior Analyst at European Schoolnet, drew attention to a dimension that underpins all AI use: data. "There is no AI without data," she noted, describing data as "the oxygen of AI." She introduced key policy recommendations from meaningful use of data in education from the Agile EDU project, which examines how educational data is collected, governed, interpreted and shared and how these choices determine whether AI systems support equity, inclusion and trust.

From evidence to national strategies

European schoolnet launched the findings of its latest report Artificial intelligence in school education, an overview of how 23 European education systems are approaching AI in schools. It maps existing national policies and guidance, the status of generative AI, how AI literacy is being integrated into curricula, and the types of training and support offered to teachers and school leaders. It also highlights examples of AI use in schools, ongoing pilot projects, and countries' main priorities and support needs for the short and medium term.

Across Europe, ministries of education have moved beyond early pilots towards more structured responses to AI: national strategies, ethical guidance, professional development for teachers and curriculum reforms are now emerging in many systems. Yet this progress has also exposed fragmentation, uneven readiness and persistent questions around data governance, assessment, teacher capacity and student wellbeing.

The second plenary roundtable turned attention to how these challenges are being addressed at national level.

Speaking on behalf of the host country, Mónica Domínguez García, General Director of Evaluation and Territorial Cooperation at the Ministry of Education, Vocational Training and Sports of Spain, echoed this cautious but determined stance. Spain's objective, she stressed, is clear: "AI should strengthen public education, not replace it." She highlighted Spain's investment in critical digital literacy, large-scale teacher training and the recently published national guide on the use of AI in education, designed to provide practical and ethical orientation for schools, teachers, students and families.

Contributions from Estonia, Luxembourg, Slovakia and Spain illustrated diverse approaches shaped by different governance structures and educational traditions.

Despite these differences, strong commonalities emerged. Daniela Hau, Head of Innovation at SCRIPT (Luxembourg) described how national strategies must balance innovation with legal and ethical safeguards, insisting that "pedagogy must come before technology." From Estonia, Kristina Kallas, Estonian Minister of Education and Research, highlighted an ambitious teacher-first approach, with extensive professional development seen as the foundation for any system-wide change. In Slovakia,  Michal Rybár, Director of Department for Digital Transformation, Ministry of Education, Research, Development and Youth, Slovak Republic shared that the focus is on short, adaptive strategies reflected in the fast-moving nature of AI, while  Julio Albalad Gimeno, Director of the National Institute of Educational Technologies and Teacher Training (Spain), emphasised the importance of coordination and guidance.

Throughout the discussion, it became clear that strategies alone are not enough if teachers are expected to navigate AI by themselves. Participants pointed instead to the importance of practical support, learning from peers and long-term investment in professional development.

Taking stock: a shared European moment

This national perspective was firmly anchored in a wider European vision. In a video message, Pia Ahrenkilde Hansen, Director-General for Education, Youth, Sport and Culture at the European Commission, underlined the urgency of the debate. With more than 80% of European teenagers already using AI regularly for learning, and teachers increasingly exposed to AI tools, she noted that "the question is no longer whether AI will be used in education, but when and how it makes sense to use it responsibly." She outlined forthcoming European initiatives, including updated guidance on the ethical use of AI and data, the AI literacy framework developed jointly with the OECD, and the next roadmap for digital education and skills towards 2030.

This European framing was reinforced at the opening of the second day by Rehana Schwinninger-Ladak, Head of Unit for Interactive Technologies, Digital for Culture and Education at the European Commission. She noted that AI and generative AI in particular is now pervasive and calls for continuous critical examination, especially in education. She stressed that education systems must actively accompany teachers and learners, rather than leaving them to navigate these changes alone.

Situating this challenge within the EU's wider policy agenda, she underlined that digital and AI skills are a strategic European priority, closely linked to competitiveness, resilience and technological sovereignty. She referred to concrete delivery mechanisms currently being developed at EU level, including the Digital Decade policy framework, the Union of Skills, and the creation of Digital Skills Academies notably an AI Digital Skills Academy, alongside academies focusing on areas such as quantum technologies, and virtual worlds.

Empowering educators in the age of AI

Day two placed strong emphasis on the role of teachers. In a roundtable moderated by Benjamin Hertz, Senior pedagogical manager at European Schoolnet, speakers explored what it means to empower educators in an AI-rich environment.

Jo Tondeur, Professor at Vrije Universiteit Brussel, highlighted that empowerment goes beyond technical skills. Teachers, he argued, need the capacity to critically assess AI tools, to design meaningful learning activities, and to decide when not to use technology. Pablo Garaizar, lecturer at University of Deusto, warned of the risk of deskilling if AI is adopted uncritically, while Alain Thillay, Seconded National Expert (SNE) by the French Ministry of Education to the European Commission stressed the importance of clarity, structure and policy coherence to support teachers in navigating grey areas.

Together, their interventions reinforced a central conclusion of the conference: AI only makes sense in schools if teachers remain at the centre as professionals, designers and decision-makers.

Bringing the classroom back in

The final plenary discussion returned to practice, with teachers from Spanish schools sharing concrete examples of AI use in classrooms. Their stories reflected experimentation, creativity and caution in equal measure, and illustrated how context, age and subject matter shape what meaningful use looks like.

Rather than offering universal solutions, these examples echoed earlier reflections from both keynotes: learning is social and deeply contextual. AI can support this process but only when guided by pedagogical intent and human judgement.

From dialogue to direction

Eminent 2025 did not aim to deliver definitive answers. Instead, it offered a shared European space to confront complexity, align perspectives and connect ongoing initiatives.

Through evidence such as the latest European schoolnet report on AI in school education, the AgileEDU  resource pack, and upcoming exchanges including the upcoming European Schoolnet Academy seminar "Assessing teaching learning in the Time of AI", European Schoolnet is positioning this dialogue as an ongoing process rather than a one-off event.

As participants departed Madrid, one conclusion stood out clearly: AI makes sense in schools only when it is grounded in evidence, guided by educators, governed responsibly and aligned with the broader purpose of education. The work now continues in classrooms, ministries and collaborative European initiatives to turn this shared understanding into sustained action.