AI between regulation, new tools and questions about the value of knowledge

In the US, one of the most talked-about news stories of recent days has been the debate surrounding the Atom web application, which was launched by Arizona State University (ASU). For a monthly subscription fee of $5, the app allows users to create any number of personalised learning modules across various disciplines – ranging from project management to investment and property – using materials provided by ASU lecturers via the UI. There has been a significant backlash from academic staff, as many lecturers were not informed that their videos, slides and learning assignments were being used to compile these modules. This case raises questions regarding copyright, lecturers’ consent, attribution and transparency in the use of university materials as a source for AI solutions. Such issues are not limited to ASU, but are also becoming relevant for other universities that will be developing their own AI platforms.

A recent Quinnipiac University public opinion survey shows that the public is not ignoring these dilemmas. As many as 74 per cent of respondents believe it is important (of whom 42 per cent consider it very important) for students at universities to learn how to use AI. At the same time, however, Americans are distinctly cautious about the role of AI in key institutional processes, with as many as 59 per cent opposing the use of AI in the processing of new student applications and 52 per cent opposing its use in tutoring. It is particularly interesting to note that as many as 47 per cent of respondents believe that students are most likely to use AI to avoid studying, whilst only 42 per cent believe it helps them learn more easily.

On the same question of how AI is transforming the field of education and careers, reference is made to a study by the Lumina Foundation and Gallup, summarised in an Associated Press article dated 27 April 206. Almost half (47 per cent) of currently enrolled students in the US have ‘strongly’ or ‘moderately’ considered changing their degree programme due to concerns related to AI, whilst around 16 per cent have actually switched programmes. Interestingly, no one knows exactly which programmes are actually AI-proof, but experts advise students to choose fields they enjoy whilst simultaneously developing AI skills, as the labour market is changing rapidly. For universities, this means a need for guidance, clear communication about the competences offered by programmes, and flexibility within degree programmes.

In the same discussion, the academic perspective on the issue of teaching capabilities is significant. In an article, the Times Higher Education portal highlights how the full implementation of the European AI Act from 2 August 2026 will force universities to change virtually everything concerning the use of AI in teaching and research. Thomas Jørgensen of the European University Association (EUA) warns that the practice of lecturers using general-purpose commercial tools, such as ChatGPT, to assess student work is particularly risky, as such uses are likely to fall into the category of high-risk systems. It is expected that universities will gradually have to build their own AI infrastructure or rely on public European solutions, such as EuroLLM. The Commission’s guidelines for implementing the Act in education are not yet finalised.

The research landscape is also showing a significant shift. Stanford HAI has awarded $2.17 million in grants to 29 new research teams in fields ranging from collaborative coding and more efficient model deployment to health monitoring and the social sciences. This year’s call for proposals attracted a record 272 applicant teams, indicating that academic interest in AI has expanded into fields that were previously only tangentially touched upon by AI (e.g. religion, social sciences, education). For universities, this represents a significant shift, as funding bodies are increasingly rewarding interdisciplinary, ethically sensitive and practice-oriented AI projects, rather than merely the engineering development of models.

A similar shift is evident in an April report by the Higher Education Policy Institute (HEPI), which summarises the key findings of the 2026 Annual Survey on the Use of Generative AI Among Students. It highlights a trend that can also be applied to the Slovenian context: 95 per cent of students say that AI skills are essential to their success, whilst the proportion of those using AI directly to generate text has fallen from 64 per cent in 2025 to 56 per cent in 2026. This indicates that students are using AI in more nuanced ways (summarising, brainstorming, editing), which is also an important pedagogical message, as it opens up scope for the thoughtful integration of AI into assignments, rather than merely banning or monitoring its use.

The picture of how AI influences young people’s choice of study and career decisions is complemented by the latest EAB survey, as summarised by Inside Higher Ed. Among the 9,516 young people surveyed who are eligible to enrol in the autumn of 2025, as many as 42 per cent believe that AI will influence their choice of career path, whilst 10 per cent have already changed their field of study because of AI. The emotional dimension of the responses is particularly significant, with young people expressing ‘uncertainty’, ‘concern’, ‘nervousness’ and even ‘despondency’ regarding the impact of AI on their future. A third of those surveyed believe that, as a result of AI, fewer professions will require a university education, which also directly affects how they view the value of higher education itself. From this, we can identify two key tasks for universities: it is essential to strengthen career advice and vocational guidance in the context of AI, and to demonstrate transparently to students how degree programmes develop the skills that will remain valuable in the labour market in the years to come.

At European level, after around 12 hours, negotiations in the final trilogue on the so-called AI Omnibus (a proposed legislative package intended to amend the AI Act) broke down. No political agreement was reached, and the main outstanding issue remains the conformity assessment framework for high-risk AI systems embedded in sector-regulated products (e.g. industrial machinery, medical devices). An important consequence for universities and research institutions is that the deadline of 2 August 2026 for the full application of obligations relating to high-risk AI systems remains in force, as without the adoption of the Omnibus, the original dates set out in Regulation 2024/1689 apply. The next political trilogue is scheduled for around 13 May 2026. In addition, in April, the Council of Europe’s Committee on Human Rights presented a handbook on the application of human rights standards to AI in the public sector (including education and social services), which is likely to influence future institutional policies at universities. Further details can be found in the April CDT Europe newsletter.

Alongside regulatory shifts, attention should also be paid to the geographical shifts in AI talent. Euronews reports on a new study which suggests that Europe is becoming one of the key destinations for AI professionals, although the US and India still lead the way. With 4.19 AI professionals per 1,000 inhabitants, Ireland is the second-largest market for AI talent in the world (just behind Singapore), Germany ranks fourth globally with 17,000 AI engineers, whilst the Netherlands is attracting American specialists who are relocating to Europe due to stricter immigration policies. For higher education institutions in the region, this has two implications: it presents an opportunity to recruit researchers and mentors, whilst at the same time increasing competitive pressure to retain their own staff.

Fortune’s article paints a contrasting picture, pointing out that whilst Europe has the talent and funding to compete with the US in AI, it is too dependent on the so-called ‘Magnificent Seven’ (the seven largest US technology companies). The author proposes the reallocation of venture capital and the development of European computing capacities as a prerequisite for ensuring that model training and research take place at home, not just in the US. This is important for the academic sector because it directly influences which computing infrastructures and models (commercial or open-source, American or European) will be available for teaching and research purposes.

Two significant developments have taken place in Slovenia in the last few days of April. The first is substantive. An editorial entitled The Illusion of Competence in the Age of Artificial Intelligence was published on the Delo.si portal, highlighting a highly topical educational challenge: where AI assists in the production of linguistically and stylistically polished outputs, it can create the illusion that knowledge, understanding and competence are present, even though in reality they are lacking. The author points out that this is not merely a problem for students, but increasingly also for institutions that use AI to prepare reports, analyses and communications. For higher education, this means that a clear distinction must be made between the student’s work and their competence when assessing and developing assignments, a point confirmed by the aforementioned international shifts towards oral presentations and process-based assessment.

The second shift is institutional and cultural. The Kresija Gallery in Ljubljana hosted, until 3 May, the exhibition ‘The Origin of the Future: Five Decades of AI Development in Slovenia’, organised by the Computer Museum in collaboration with the Faculty of Computer and Information Science at the University of Ljubljana, Slo AI Institute, the Slovenian Association for Artificial Intelligence (SLAIS) and the Jožef Stefan Institute. The exhibition demonstrated how, as early as 1970, the first researchers at the Jožef Stefan Institute began to break new ground in a field that is now a central concern of science, education and society. At the end of April, guided tours and holiday workshops were organised as part of the exhibition, and on the final day of the exhibition, 3 May 2026, two guided tours in English took place. The exhibition serves as a reminder that Slovenia is building on five decades of its own research tradition in the field of AI, and that this historical perspective is important in shaping current institutional policies and curricular decisions.

It is also worth adding a small practical note. As reported by the Delo.si portal, Anthropic’s new tool, Claude Mythos, is so powerful at identifying security vulnerabilities that the company is not currently permitting its release for general use, whilst the Slovenian Information Security Agency is warning organisations of an increased cyber risk. For university IT departments and research teams, this serves as a reminder that, when introducing AI, it is essential to also consider security and ethical assessments of its use, not just its educational and research benefits.

Developments in the second half of April indicate that, within the higher education sector, AI is no longer merely a matter of considering which tools to use for which tasks, but rather raises a systemic question concerning three intertwined dimensions: namely, rules (ranging from European regulations to educators’ consent for the use of their materials), the value of knowledge (how to prevent the illusion of competence and ensure genuine student work) and infrastructure (who will own the computing resources, models and data used by universities). The thoughtful integration of AI into university processes thus increasingly requires collaboration between teaching staff, IT departments, university management and the wider regulatory environment, as individual decisions quickly give rise to systemic consequences.

If you have tried out an interesting use of AI in education at your faculty, as part of a course or within the context of your professional work, we warmly invite you to share your example with us via the form for collecting examples of good practice. We look forward to reviewing the submissions and including selected contributions in future issues of the AI newsletter, so that together we can build an overview of useful and well-considered approaches within the university environment.

Authors: Eva Kern Nanut, Eva Škraba, Maja Kosmač, Sanja Jedrinović Čufer, Mateja Bevčič, University of Ljubljana Centre for the Use of ICT in the Educational Process