Artificial Intelligence in Education and Science: Between Opportunities, Caution, and Strategic Considerations

An international conference on artificial intelligence in education was held in Maribor earlier this month, bringing together more than 100 teachers from 45 countries. The conference was co-organized by the Asia-Europe Foundation (ASEF), the International Research Centre On Artificial Intelligence (IRCAI) and its Open Education for a Better World program, and the Anton Martin Slomšek Diocesan High School in Maribor, with support from the Ministry of Foreign and European Affairs, the National Commission for UNESCO, and other partners. Teachers from European and Asian countries worked in mixed groups to develop innovative examples of how to use artificial intelligence in educational settings. “This is about an international perspective; we’re not confined solely to our own context or our own classroom. Artificial intelligence knows no boundaries, and broader perspectives are undoubtedly important—education, too, is becoming increasingly global,” says Mateja Brejc of the Ministry of Education. They were looking for answers to questions about how artificial intelligence can improve learning processes while remaining accessible, inclusive, and ethical (more information).

The third working conference on artificial intelligence and education, titled “Ensuring Quality Education in the AI Era – Introducing the Council of Europe Compass for AI and Education,” organized by the Council of Europe, brought together policymakers, educators, academics, experts from the technology sector, and civil society to discuss the future role of AI in education. The discussions highlighted the need for a common framework that would prevent fragmented standards, inequalities, and excessive commercial influence, while simultaneously fostering trust, innovation, and the cross-border recognition of qualifications. Among the challenges mentioned were the digital and generational divides, adapting the roles of teachers, and ensuring adequate resources for administration. The development of AI literacy should be based on digital and media literacy and promote critical thinking and civil participation; however, challenges remain regarding unequal access, teacher burnout, overloaded curricula, and limited funding. Legal discussions have emphasized a human-centered approach grounded in human rights, with existing legislation helping to mitigate risks. Key themes have shown that Europe does not seek to control technology, but rather to guide its use through shared values, ensuring that education remains public and democratic. Trust, transparency, cross-sector collaboration, and the involvement of teachers and students are key to meaningful innovation (more information).

An article we came across on the website of the British institute HEPI, titled “Slowing Down AI in Higher Education,” points out that, despite the general enthusiasm for artificial intelligence, a measured approach is needed. It introduces the concept of “Slow AI” as a way to thoughtfully integrate technology into the educational process. The article advocates for the reflective use of tools that encourage students to reflect on their own learning rather than to automate their thought processes. The advantage of this approach is that it preserves the authenticity of education and fosters the development of critical thinking; the disadvantage is the risk that development and innovation may slow down at a time when the world is changing rapidly. The challenge raised by the article is striking a balance between responsibility and progress, and determining how to introduce technology thoughtfully without falling behind (more information).

From a completely different perspective on artificial intelligence, a report by the European Commission’s Joint Research Center (JRC) titled “A European Strategy for Artificial Intelligence in Science: Paving the Way for the Resource for AI Science in Europe (RAISE),” states that artificial intelligence has become a key tool for accelerating scientific research and strengthening European competitiveness. The document presents the new RAISE (Resource for AI Science in Europe) strategy, which, among other things, emphasizes the need for shared infrastructures and open science to ensure reproducibility, broader access, and reliability. While artificial intelligence models are becoming increasingly powerful and flexible, they also require significant resources for training and deployment. For this reason, investments in high-performance computing, AI factories (ecosystems that foster innovation and collaboration), and open scientific databases are essential to ensuring the EU’s leading position in AI research. Greater use of artificial intelligence in science would create new demands for specialized expertise among researchers. The JRC’s assessment of skills and expertise points to the need for “hybrid” (multidisciplinary and interdisciplinary) teams that combine expertise in engineering, computer science, and artificial intelligence with domain-specific expertise. Policies should therefore focus on attracting, developing, and retaining this interdisciplinary talent. The advantage of such an approach is enhanced collaboration and greater efficiency in research processes, while the disadvantage is dependence on complex infrastructure systems and potential inequalities in access across research environments. The challenge arising from this strategy is how to ensure that the benefits of artificial intelligence in science are accessible to all, not just the largest institutions (more information).

An article published on the European Commission’s website under the title “Keeping European Industry and Science at the Forefront of AI” continues this strategic discourse. The Commission is announcing investments worth several billion euros to develop artificial intelligence in industry and science, with the aim of maintaining European technological sovereignty. The article outlines an ambitious plan to bring together research and industry partners and to accelerate the development of applications that will move from the laboratory to practical use. The advantage of this approach is systemic support and the strengthening of Europe’s role in global competition; the disadvantage is the risk that technological development will outpace social and ethical considerations. The challenge for Europe is how to remain a leader in the development of artificial intelligence while maintaining human oversight, transparency, and security (more information).

OpenAI unveiled Sora 2, a new model for generating video and audio, along with a social app for iOS featuring safety features. Users quickly began creating copyrighted characters, exposing gaps in the framework for responsible use and forcing OpenAI to implement a series of changes. Updates to Sora 2 include the ability for copyright holders to access detailed on/off controls that allow them to manage how users create their characters, monetization of videos through revenue sharing with rights holders; and expected frequent updates, as OpenAI tests various approaches within the Sora model before rolling them out to other products (more information).

Starting December 16, 2025, Meta will begin using people’s conversations with its AI chatbot to personalize ads and content, and users will not have the option to turn this off. This marks a significant shift in how personal interactions with the AI are being turned into monetization tools, meaning that casual questions—such as a request for hiking recommendations—could directly trigger targeted advertising for related products (more information).

Teach For America’s Reinvention Lab engaged educators in a three-stage artificial intelligence system (hackathons, prototyping workshops, and Arcade AI) and showed them a practical path to “AI literacy.” It positioned teachers as creators who amplify impact by imparting AI knowledge to students—knowledge that is essential for their economic mobility. More than 1,200 educators actively participated in AI-related training experiences, during which more than 720 teaching tools were created, and more than 10,000 students benefited from AI solutions developed by the educators (more information).

Budget Lab analyzed the impact of artificial intelligence on the U.S. labor market since the introduction of ChatGPT and found no measurable disruptions in employment or occupational patterns. The occupational structure is changing slightly faster than during previous technological transitions (about 1% faster than during the Internet era), but these trends began before the era of artificial intelligence. Current measures of exposure to and use of artificial intelligence show no correlation with changes in employment or unemployment rates. The study acknowledges significant data limitations, including reliance on theoretical measures of exposure and usage data from only one AI model, and emphasizes the need for comprehensive, transparent data from all major AI developers (more information).

Software and tools for detecting the use of generative artificial intelligence, including paraphrasing detection, have become increasingly widespread in education as institutions respond to the challenges posed by AI-generated content. The authors of a study titled “Heads we win, tails you lose: AI detectors in education” point out that these detectors cannot effectively identify AI-generated content because they do not allow for verification of the text’s true origin. Attempts to test these tools under real-world conditions have shown that the results are often unreliable and can lead to false conclusions. Furthermore, the study highlights the challenges of implementing these technologies in educational systems, as they can result in false accusations of academic dishonesty, which harms students. The authors suggest that ethical considerations be taken into account when introducing AI detectors and that more reliable and fair approaches to assessing student work be developed (more information).

As you may have noticed, we’ve had two eventful weeks in the field of artificial intelligence. The common thread running through all these articles is the realization that artificial intelligence is no longer a matter for the future, but for the present. Universities, research centers, and political institutions are all facing the same question: how to design infrastructures, practices, and values that will integrate artificial intelligence into the educational and scientific processes in a way that enhances knowledge rather than diminishes understanding. The opportunities are vast, ranging from personalized learning to accelerated research, but the challenges are equally great: ethical frameworks, data security, transparency, and responsible use.
In the coming months and years, it will be crucial for higher education institutions not to settle for merely the technical implementation of artificial intelligence, but to develop their own culture of understanding and critical use of it. Only in this way can we ensure that artificial intelligence not only accelerates processes but also deepens their substance, and that it remains at the service of humanity—not the other way around.

Authors: Eva Škraba, Sanja Jedrinović Čufer, Mateja Bevčič, Eva Kern Nanut, University of Ljubljana Center for the use of ICT in pedagogical process