Today the OECD released the results of PISA 2025. They make uncomfortable reading. The results point to a broader challenge for education systems, including Sweden, and raise questions about whether young people are developing the fundamental competencies in mathematics, reading and science that they will need in a rapidly changing world.

At the same time, another transformation is taking place at extraordinary speed. AI has already become part of how young people learn and how we work at universities.

I think we need to consider these two developments together.

Much of the discussion about AI in universities has concerned what AI can do. How can it improve teaching? How should students be allowed to use it? Can it accelerate research? These are important questions, and at SCI we have been working actively with them in our guidance for AI in education. But there is another question that I increasingly find more fundamental:

What does it mean to become scientifically competent in the AI generation?

A student can ask AI to explain a theorem, solve an equation, write code or summarise a scientific paper. A doctoral student can use it to search literature, suggest hypotheses, analyse data or propose steps in a proof. Researchers are already experimenting with systems that automate increasingly large parts of the scientific process.  This is an extraordinary opportunity for science. But it creates a paradox:

The easier it becomes to produce an answer, the more important it becomes to know how to judge the answer.

We learn to become researchers partly through struggle. A doctoral student searches the literature, writes code that does not work, follows an argument before finding the mistake, writes and rewrites, and discusses half-formed ideas with a supervisor. AI can remove much of this struggle, and some of it we should be delighted to remove. Science has always progressed by developing tools that allow us to stop doing things that machines can do better.

But some of this struggle may also be where scientific judgement is formed.

This concern is beginning to appear explicitly in the scientific discussion: AI may accelerate scientific production while narrowing scientific inquiry and affecting how expertise and judgement develop ( Nature comment, “The uncritical adoption of AI in science is alarming — we urgently need guard rails“)

This has consequences for both doctoral education and the role of the professor. If AI can perform some of the activities through which we traditionally learned to become researchers, we need to understand what can safely be delegated and what is essential to developing an independent scientist.

At the same time, AI is narrowing the traditional gap in access to knowledge and technical expertise between professor and doctoral student. What experience provides, however, may be something harder to reproduce. Which of ten plausible directions is worth pursuing? When should we abandon an approach? What result would actually change our understanding? When is an unexpected result a mistake, and when is it the beginning of something interesting?

Perhaps the professor of the AI generation will become somewhat less important as a provider of answers and even more important in helping young researchers learn how to ask questions and exercise judgement.

And this brings me back to PISA 2025. The students who will enter KTH in a few years are growing up in an environment where access to sophisticated answers is becoming almost unlimited, while we are seeing worrying signals about fundamental competencies in education.

The response cannot be to protect students from AI, nor simply to teach them to use it more efficiently. Generative AI can support learning, but should enrich it rather than replace cognitive effort or teachers’ professional judgement (The OECD Digital Education Outlook 2026)

Our responsibility as educators is therefore larger: it involves fostering curiosity, persistence and the confidence to question what a machine or, for that matter, a professor tells you. These are skills that have always been part of education, but have perhaps never been as fundamental as they are today.

The AI generation will need more scientific competence, not less.

Our challenge as universities is to make sure they acquire it.

Text developed in collaboration with AI.

How do we develop education, research environments and academic culture in a time of rapid technological and societal change? At the School of Engineering Sciences, SCI, we are working actively with three focus areas that are central to this question: AI in education, hands-on and project-based learning through SCI MakerSpace, and inclusion as a foundation for a strong academic environment.

To make this work more visible and easier to follow, we have now published three dedicated pages:

• AI for education
• SCI MakerSpace
• Inkludering@SCI

These pages gather information about ongoing activities, workshops, resources and contact points. They are meant to be living entry points into work that will continue to develop together with faculty, staff and students.
The three areas are different, but closely connected. AI is already changing the conditions for learning, teaching, examination, administration and research support. MakerSpace will create new opportunities for students and teachers to design, build, test and develop ideas in a practical and creative environment. Inclusion concerns the academic culture that makes it possible for people to contribute, collaborate and grow.

A central part of our approach is direct involvement. These focus areas are not being developed only through strategy documents or central decisions, but through workshops, discussions, specialized meetings and concrete activities involving faculty, staff and students. Lasting change in education and academic culture requires shared understanding, local engagement and opportunities for colleagues to bring in their expertise and experience.

A recent example was the full-day Agentathon workshop on AI agents, in which I also took part. It was an impressive source of inspiration, and a very tangible sign of both the opportunities and the risks connected to AI. We saw how AI tools can make many daily tasks faster, more effective and more structured. At the same time, the workshop raised serious questions about security, responsibility, data handling and the need for critical judgement.
This balance is important. Curiosity and experimentation are necessary if we want to understand how new tools can strengthen our work. But they must go together with awareness of risks, ethical considerations and sound academic judgement.

The publication of these pages is therefore not an endpoint, but an invitation to follow, contribute to and discuss the work as it develops. Each of these focus areas touches questions that are central to the future of universities: how we use new technologies responsibly, how we create environments for creative and practical learning, and how we build academic cultures where people can contribute fully. By making the work visible and by developing it together with faculty, staff and students, we strengthen not only individual initiatives, but also the conditions for education, research and collaboration at SCI.

Each spring, the first-choice application numbers give us an early signal about the next cohort of students. They are not the whole story – final admissions depend on capacity, eligibility, selection groups and final choices – but they are a useful indicator of attractiveness and long-term positioning.

SCI: a stable portfolio, with a clear upward signal in two programmes

For SCI, the overall picture is stability – paired with a very clear upward signal in two of our flagship programmes. In the latest application round (first-choice applicants), the SCI programmes show the following snapshot:

• Engineering Physics (Teknisk fysik): 534 → 672
• Vehicle Engineering (Farkostteknik): 287 → 363
• Engineering Mathematics (Teknisk matematik): 288 → 262
• Open Entry (Öppen ingång, SCI): 264 → 248

What stands out is the magnitude of the increases in Engineering Physics and Vehicle Engineering, both around +26% year-on-year. The remaining programmes show the kind of modest year-to-year movement we typically see in application cycles. Taken together, this points to a stable demand for SCI’s educational portfolio, with particularly strong momentum in programmes that offer a broad engineering foundation and multiple pathways for specialization.

One possible interpretation (offered cautiously!) is that broad programmes with flexible pathways are becoming even more attractive. In a labour market where AI is changing tasks and skills, many students may prefer degrees that keep options open: strong fundamentals, flexibility in specialization, and room to connect theory with emerging technologies over the course of the programme.

Are young people changing their study choices because of AI?

A common question right now is whether the rapid spread of AI is already reshaping how young people choose what to study.
The national numbers suggest that interest in engineering is not weakening. UKÄ reports that civilingenjör programmes had 16,500 eligible first-choice applicants for autumn 2025, an 8% increase compared with autumn 2024. UHR’s admissions statistics also show broad continued demand: for autumn 2026, applications increased to both courses (+2%) and programmes (+5%).
At the same time, it is clear that many students expect AI to affect their future working lives. In an OECD survey on experiences with generative AI, over 75% of respondents under 35 describe AI as useful, and many in the same age group expect AI to have a substantial impact  on their careers. This matches what we hear informally as well: students increasingly ask not only “what am I interested in?” but also “what will still matter in five to ten years?”

The labour-market outlook from major international analyses points to restructuring rather than a simple job-loss narrative. The World Economic Forum’s Future of Jobs Report 2025 projects 170 million jobs created and 92 million displaced globally over 2025–2030, a net increase of 78 million jobs, while emphasizing that skills needs will shift rapidly. The European Commission similarly notes that AI and emerging digital technologies are expected to bring productivity gains and can increase employment overall, but with uneven effects and clear needs for reskilling.
In that context, SCI’s stability, and the strong increase in programmes with broad foundations, fits a plausible pattern: students are not avoiding engineering because of AI. Instead, many appear to be seeking programmes that combine rigorous fundamentals with adaptability.

Numbers are not a goal in themselves. The real question is what we do with the opportunity.
As we prepare for the incoming students of 2027, SCI will continue to strengthen educational quality and student experience, by focusing on two forward-looking developments:
    • AI in education: building the right competence among faculty and supporting students in using AI responsibly and effectively, in ways that deepen learning and uphold academic integrity.
    • Makerspace integration: strengthening hands-on, prototype-oriented learning environments where students can meet across programmes and connect with researchers, bringing theory closer to experimentation, iteration, and engineering practice.

Congratulations to KTH, and to the other Swedish universities that have secured new resources through the Strategic Research Areas initiative. For SCI, this is particularly meaningful through our engagement in quantum technology and polar research.

At the same time, the emerging VR-Vinnova excellence-cluster initiatives point in a similar direction: towards stronger and more visible research environments built through collaboration across disciplines, institutions, and sectors.

This also brings to the surface an issue that is not always discussed openly: coordination has a cost.

In academia, we rightly value academic freedom. The freedom to define our own questions and pursue our own ideas is fundamental to research. At the same time, major strategic initiatives are built on a broader logic. They aim to create environments that are not only excellent in parts, but strong, coherent, and internationally visible as a whole.

That requires more than outstanding individual researchers. It requires cooperation across boundaries, the willingness to build consolidated groups, and sometimes also the readiness to engage in efforts that go beyond one’s own immediate research interests or local priorities.

This is not always easy, nor should it be taken for granted. Coordination takes time, attention, and flexibility. It can create friction, and if handled poorly, it can easily become more process than purpose.

But when important opportunities arise, especially when substantial resources are connected to broader national priorities, it becomes essential to ask not only what each of us wants to pursue, but also what we can build together.

That, perhaps, is the real challenge. Not to weaken academic freedom, but to complement it with a culture of cooperation strong enough to meet ambitions that are larger than any one researcher, group, or institution. In that sense, alignment should not be understood as uniformity. It is better understood as a collective capacity: the ability to gather strong competences around shared goals when the moment calls for it.

As many new alliances are now taking shape, this seems particularly timely. The question is not whether coordination is desirable in itself, but when it creates real value, and whether we are prepared to invest in it when it matters. If we get that balance right, coordination becomes more than an administrative necessity. It becomes part of how strong research environments are built.

I am pleased to share SCI’s Verksamhetsplan 2026. A plan is not a goal itself, its value is whether it helps us focus, coordinate, and follow through. SCI is a large and diverse school, with world-leading research environments, ambitious educational programmes, and a community that spans many disciplines and backgrounds. In that setting, clarity matters: we need shared priorities that help us make good choices, and we need a common language for how we evaluate progress over the year.

For 2026, I want to highlight three main focus points:

1) AI in education: competence and shared practice.
Generative AI is already part of our students’ learning reality and of many colleagues’ teaching practices. Our task is to ensure that AI becomes a resource that supports learning, not a shortcut that undermines it. In 2026 we will strengthen competence for both faculty and students and, just as importantly, facilitate structured sharing of experiences, teaching materials, and course experiments across SCI so we learn faster together.

2) Makerspace: a place to build, test, and meet.
As an engineering school, we benefit enormously from environments where ideas can move from concept to prototype. A Makerspace is not only about tools; it is about creating a shared arena where students and researchers can meet across programmes and disciplines, build prototypes, test ideas, and learn by doing. Developing this environment is a concrete way to strengthen both education and innovation culture at SCI.

3) Quality and inclusion: the foundation of long-term strength.
Quality in education and research is inseparable from the environment in which we work and study. In 2026 we continue to embed quality, equality, diversity, and inclusion in everyday leadership and decision-making, particularly in recruitment, competence supply, and how we support colleagues and students in their development. A sustainable work and study environment is not an “extra”; it is a prerequisite for excellence.

We will also continue the transparent reporting of financial resources that we have built over the last years. It has become an important part of how we create shared understanding and enable good planning at all levels.

I encourage you to read the Verksamhetsplan 2026 as our shared direction for the year ahead, and as an invitation to dialogue. The plan will guide priorities at school level, but it is only through the work in departments, units, programmes, and research groups that it becomes real. Thank you to everyone who contributed to shaping it, and thank you in advance for the collective effort that will turn priorities into outcomes.