Education & Empowerment
The dual system is the greatest structural asset. AI competencies must enter the curricula now.
Where does Switzerland stand, and where are the fault lines?
About two thirds of young people in Switzerland complete initial vocational education and training. It is governed through the proven partnership of the Confederation, cantons and labour-market organisations, under the leadership of the SERI. This dual system is closely aligned with the labour market and responds to new demands more quickly than purely school-based models. At the top sit ETH Zurich and EPFL, which demonstrate through the open language model Apertus that Switzerland can build independent AI competence. At the same time, adult continuing education remains fragmented, voluntary and unevenly distributed, despite the Federal Act on Continuing Education and Training. With its dual education system Switzerland holds the best instrument for adaptability, yet has not fully used it to build AI competence.
AI is changing occupational profiles faster than curricula normally follow. The strength of the dual system is its close ties to today's firms and occupations. That strength can become a weakness if those occupations change rapidly. What counts is whether initial training, universities and continuing education together convey AI competence, for everyone, not only for those already well qualified. Will lifelong learning become a reliable infrastructure, or remain a fragmented offer for those who need it least?
Pathways for Switzerland
Switzerland can build AI competence systematically into compulsory schooling and initial vocational training, from critical engagement with AI tools through to their use in the workplace. The joint-partner structure of the dual system allows training regulations to be updated comparatively quickly, because businesses sit at the table themselves. This works well because structures and governance channels already exist and the need is undisputed. If AI competence is embedded early and broadly, it becomes standard equipment rather than a privilege. Constraints lie in the pace of curriculum revisions and in the continuing professional development of teachers.
Trade-off: broad foundational competence for all versus the effort of continuously updating curricula and teaching staff.
Feasibility: Switzerland's greatest lever. This is where its structurally superior education system has the most direct effect.
Adult continuing education in Switzerland is heavily fragmented and depends on individual initiative and employers. Individual learning accounts or training credits would give every person a reliable budget for lifelong learning, independent of their employer or prior qualifications. This matters because those with lower formal qualifications participate least in continuing education, even though they are most exposed to AI-driven change. Such an entitlement would underpin the Federal Act on Continuing Education and Training with a concrete individual right. Funding and the federal division of responsibilities between the Confederation and the cantons remain the main challenges.
Trade-off: a universal entitlement to learning for all versus considerable costs and administrative complexity.
Feasibility: Closes the largest gap in the Swiss system: unequal access to continuing education.
With ETH, EPFL, the Swiss National Supercomputing Centre CSCS and the universities of applied sciences, Switzerland has the foundations to build sovereign AI capability rather than relying solely on foreign providers. The open language model Apertus shows this is achievable. That capacity could be developed deliberately as national infrastructure for research, education and public applications. This matters because digital sovereignty and capability are linked. Those who can build and understand AI themselves are less dependent. Expanding this infrastructure requires substantial funding and specialists, both of which are already scarce.
Trade-off: digital sovereignty and cutting-edge expertise versus substantial investment and resource commitment.
Feasibility: A pillar of Switzerland as a business location. It combines research strength with independence from foreign platforms.
Higher vocational education — with its federal examinations and higher specialist colleges — represents a distinctly Swiss pathway for requalifying people mid-career. Targeting it towards AI-related competencies would open advancement for those whose initial training is being devalued by automation. This matters because adaptation does not end at the point of entering the workforce, and Switzerland already has an established, practice-based framework to build on. Higher vocational education also enjoys strong acceptance in the labour market. What remains unresolved is funding for those who need to retrain alongside work and family commitments.
Trade-off: upskilling and requalification for adults versus significant personal and financial demands.
Feasibility: An underestimated asset: higher vocational education is a ready-made instrument for retraining.
Trigger-based phases
Trigger: AI tools are spreading through organisations faster than curricula and continuing education can absorb them.
At this stage, workers are acquiring AI skills largely on their own initiative, unevenly distributed. Initial training regulations and pilot modules are being adapted, and universities are building capacity. The pressure to act is growing, but the dual education system has the structure to catch up quickly.
Trigger: AI competence is becoming a baseline requirement across many professions, and the continuing education gap is gaining political visibility.
AI competency is embedded in vocational initial training and compulsory schooling, while individual learning accounts or an entitlement to continuing education close the gap for adults. ETH Zurich, EPFL and the universities of applied sciences function as national infrastructure, providing models, tools and continuing education. Lifelong learning becomes reliable rather than a matter of chance.
Trigger: the ability to work with and on AI has become an assumed standard capability.
Education and continuing training form a coherent system running from compulsory schooling through to adult retraining. Switzerland understands capability as a central factor in its competitive position and thereby maintains its lead. The advantage of the dual education model is fully realised and complemented by sovereign AI competency.
Real-world pilot programmes
Status: active. ETH Zürich, EPFL and CSCS, published 2025
Apertus is a fully open, transparent and multilingual language model developed by ETH Zurich, EPFL and the Swiss National Supercomputing Centre CSCS. Because its training data, weights and methodology are publicly available, it serves as a traceable foundation for research, education and public applications. The project demonstrates that Switzerland can build independent, sovereign AI capability rather than relying solely on closed foreign models. It shows concretely how research strength translates into national capability.
Status: established. The Confederation, cantons and the world of work under the leadership of the SBFI
Through the federal partnership, the Confederation, cantons and vocational and professional organisations develop training regulations jointly and close to the labour market. This allows occupational profiles and competences to be updated as requirements change. That is a structural advantage over purely school-based systems. For integrating AI competency, this framework is the decisive infrastructure. The open question is whether the pace of revisions can match the speed of AI-driven change.
Status: reference model, to be assessed as a template for Switzerland
In Singapore, every adult receives a government-funded continuing education credit that they can use independently for approved courses. The system makes lifelong learning a reliable individual entitlement rather than a matter of circumstance and employer discretion. For Switzerland, it offers a conceptual model for underpinning fragmented continuing education with individual learning accounts. Direct transfer is not feasible given Switzerland's federal structure, but it remains instructive as a template for an enforceable right to learn.
What we are watching
This indicator measures how large the gap in continuing education participation is between high- and low-qualified workers. It matters because low-qualified workers are most exposed to AI displacement yet least likely to engage in learning. If the gap widens further, inequality grows rather than shrinks. For education policy, this is the central measure of whether capability reaches everyone or only those who are already strong.
This tracks how demand for apprenticeship places is evolving, particularly in occupations with high automation risk. The indicator matters because it shows at an early stage whether young people and employers regard certain occupational profiles as viable for the future. A sustained decline in AI-exposed fields signals that initial vocational training needs to be adapted. For SERI and its trilateral partners, it is an early indicator of where revisions are required.
This tracks how many programmes include AI competency as a mandatory element. It matters because curricula otherwise fall behind workplace reality. If integration stagnates while AI spreads rapidly, the gap between training and the labour market grows. For education authorities, it is the measure of whether the system is keeping pace with change.
This measures how much computing power and how many open models Switzerland has at its disposal domestically, including at CSCS. It matters because capability and digital sovereignty are linked. Those who can build AI themselves remain independent. If dependence on closed foreign systems grows without domestic expansion, room for manoeuvre shrinks. For research and industrial policy, this is an indicator of strategic autonomy.