Training, Reskilling, Alpine Campus

Who learns what, where, and how, in a canton whose workers are about to see their trades transformed at an unprecedented pace? This question looks more modest than the one about sovereignty addressed in the previous chapter, yet it is probably even more decisive for the ten years ahead.

It does not receive, in Valais public debate, the attention it deserves. Discussion of training focuses on students: how many are enrolled in higher education, which programmes are growing, which new campuses need building. These questions matter, and the canton is working on them in earnest: HES-SO Valais-Wallis now enrols close to three thousand students, growing steadily, and three major campus projects are under way in Sion, Sierre and Brig, targeting 2030⁴³. That is a sound trajectory, one that consolidates an initial-training apparatus the canton needs.

But this discussion leaves in the blind spot the question that strikes me, over a ten-year horizon, as the most structurally important one: how to reskill workers already in the workforce. Not the young people entering the labour market — higher education already takes care of them — but the forty-five-year-old winegrowers, the fifty-year-old fiduciaries, the sixty-year-old family doctors, and alongside them the hoteliers, the municipal councillors, the caregivers, the tradespeople, who make up the real economic and social fabric of the canton and whose trades will be profoundly transformed within five or ten years. Taken together, they are the mass to which the transformation this essay describes is actually addressed. And it is their capacity to take up the new tools, to move up in skill without leaving their region or their job, that will decide which slope Valais takes.

Why the seniors rather than the young

This prioritising of the working population is not simply a matter of numbers. It follows directly from the argument built up in the preceding chapters, which must now shape training policy.

The competitive shift described in chapters 2 and 10 only plays out fully where seniors with architectural experience are steering the use of AI tools. It is this trade intelligence, accumulated through experience, that becomes, in the age of AI, the strategic lock on the new cognitive economy.

The consequence for cantonal training policy is clear-cut. The priority is not, or is no longer, mainly to train the generation entering the labour market: that generation will learn fast, by imitation and by use, and higher education is already doing its job. The priority is to train the generation that today occupies positions of architecture and decision-making within Valais's economic structures — winegrower-cellarers, partners in professional firms, doctors in charge of practices, independent hoteliers, heads of small and medium-sized businesses — so that they know how to orchestrate the new tools within their trade. Without this training of seniors, the potential for a competitive shift stays theoretical. With it, it becomes a concrete regional asset.

This priority carries a further political merit. Young people entering the labour market today learn AI at home, by playing, testing, imitating. They do not need a public scheme to encounter it: they have already encountered it. Seniors already in work, by contrast, rarely have the time, the inclination or the setting to take it up on their own. They need a strong institutional signal telling them that this transformation concerns them, that it is serious, that it warrants an investment on their part. Only a coordinated public scheme can send that signal with enough authority to be heeded.

What already exists, and what is missing

Before proposing anything, let us acknowledge what already exists. Valais is no desert when it comes to continuing education. HES-SO Valais-Wallis offers a wide range of certifying postgraduate programmes — CAS, DAS, MAS, EMBA — covering business, management, engineering, health, tourism, social work. The Idiap, in Martigny, runs seminars and thematic events whose reach extends well beyond the canton. HEP-VS in Brig handles continuing education for teachers. Professional associations — fiduciaries, doctors, engineers, winegrowers — offer their own sector-specific schemes. The chambers of commerce, the Valais economic development agency and the Fondation The Ark occasionally run awareness events.

This landscape is richer than is often perceived. But three of its features limit its capacity to carry the transformation we are discussing.

Fragmentation, first: the schemes exist, but they sit side by side rather than working together. A fiduciary wanting to understand how AI will transform their trade does not know where to turn — HES-SO, their professional association, a private provider, a course outside the canton? The offer is there, but it has no legible front door. This fragmentation particularly penalises small structures, which have neither the time nor the means to map the training market in order to find their way through it.

The diploma-driven orientation, next: Valais continuing-education programmes almost all lead to a certificate or a diploma. That is a value in itself, but this format does not match the needs of workers who, at forty or fifty, are not looking for a new qualification but for concrete ways to fold new tools quickly into their daily practice.

And finally, the essentially sector-based character: existing training, when it addresses AI at all, does so within specific tracks — AI for engineers, AI for care workers, AI for economists. Yet the transformation the canton is living through cuts across sectors. A winegrower, a fiduciary, a home-care worker, a municipal employee, a hotelier each have different things to learn, but also common ones: what a large language model is, what it does well and badly, how to phrase a useful request, how to judge the quality of an answer, how to handle data confidentiality, how to check and govern what one entrusts to it. This common core, which ought to be offered broadly to every working person in Valais at modest cost, is today covered nowhere in any systematic way.

It is in this space — between the diploma-granting and the sector-specific, between the academic and the commercial, between the grande école and in-house corporate training — that what I call the alpine campus takes up residence.

What the alpine campus is not

To head off the misunderstanding that dogs any proposal of this kind, let us first say what the alpine campus is not: not a new school of engineering or a university establishment (HES-SO Valais-Wallis and EPFL already cover that need competently, and the canton has no mandate for a third such institution), not another online course platform (dozens already exist, and producing more of them is not the bottleneck), not a consultancy dressed up as training, and not a promotional vehicle for particular tools.

What it is, instead, is a coordinated scheme, public or quasi-public, that makes a path of skills advancement legible and accessible, tailored to each Valais worker's trade, age and place of residence. It does not invent the training it offers: it assembles part of it from what already exists, commissions another part from institutions and providers capable of producing it, and builds a few modules of its own where no one else will. It plays the orchestrator's role that no one in the canton plays today.

This orchestrating posture has an important virtue: it creates no duplicate of what already exists, and it spares public resources. Building a new institution would cost several tens of millions of francs and several years before any visible effect. Building an orchestration scheme costs a fraction of that and can produce its first results within eighteen months.

A format suited to workers already in employment

How effective such a scheme proves depends as much on its format as on its content. Several principles strike me as decisive.

Short modularity, first. Not programmes lasting several months, but units running from a few hours to a few days, combinable according to need. A four-hour module on the fundamentals of large language models, a full day on data governance, a half-day on uses specific to the trade of a fiduciary or a winegrower. These units can be strung together, but each also stands on its own. A worker who attends only a single module still gets an immediate benefit from it, which is rarely true of long programmes.

The blending of in-person and remote formats, next. For workers already in employment, scattered across the whole canton, full in-person attendance is out of the question: it is incompatible with the working lives of most people. Fully remote learning is not enough either: learning digital tools works better through hands-on workshop practice than through asynchronous video. The right balance combines both, with concentrated in-person sessions, perhaps quarterly, held in several locations across the canton, and remote support between sessions. The geography of Valais, made up of scattered valleys, requires that in-person sessions be organised at several points — at minimum Sion, Sierre, Martigny, Visp and Brig — and ideally at a few local sites in the more populous valleys.

Anchoring in the trade. A generic module on AI, however good, speaks to few people. A module that draws its examples from the concrete cases of a given trade — a tax return for a fiduciary handling a small business, a procedural memo for a municipal employee, tracing the provenance of an alpine cheese for an alpine herder, diagnosing a vine disease for a winegrower — speaks to those who take it. This requires a body of sector-specific content production, carried out with the relevant professional associations, work that cannot be delegated to providers disconnected from the ground.

Bilingualism. The canton is bilingual, and the AI transformation touches French-speaking and German-speaking workers alike. A scheme covering only French-speaking Valais, or only Upper Valais, would miss its target. Producing modules in French and standard German, and, for some content where it is relevant, in Walliserdeutsch and in patois, is not a minor detail. It is a condition of consistency with the sixth chapter of this essay, and a mark of respect for the German-speaking half of the canton.

Finally, a modest cost for the learner. If the scheme's aim is to reach a hundred or a hundred and fifty thousand workers within five to ten years, it cannot be priced like an EMBA. Making it free is not necessarily the right answer: a free module is often one people do not see through to the end. But a token cost, on the order of a few tens to a few hundred francs per module, funded partly by learners and partly by public effort, is probably the right setting. It signals to the learner that the training has value, without becoming a barrier to entry.

Three tracks for three audiences

The alpine campus brings together several distinct tracks, according to audience and objective.

A shared literacy, first: the broadest track, and probably the simplest to set up. A few short modules, four to eight hours in total, laying the common groundwork for every Valais resident — what a large language model is, what it does well and badly, how to phrase a request, how to check an answer, how to handle confidentiality. This track must be accessible on a very large scale, in person and remotely, in both languages of the canton, at a token cost. Its purpose is not to train experts but to prevent most people in Valais from using these tools purely by imitation, without understanding how they work. It is an investment in collective literacy, comparable to what the cantons did for basic computer literacy at the turn of the millennium.

This is also the track best suited to deployment at municipal scale. A Valais commune of a few thousand inhabitants can, in partnership with the other communes of its district and with support from the cantonal scheme, host an annual in-person session open to all its residents, not only its workers. The bourgeoisies — the traditional Valais civic corporations that manage collective land and resources — have historically carried this kind of collective instruction, and several communes already have the pedagogical infrastructure for it: communal halls, primary schools with staggered timetables, association premises. This is the scale at which the canton's institutional grammar — that third way between state and market described in the fourth chapter — can take in the alpine campus without adding bureaucracy. And it is at this scale that the scheme becomes visible to residents who would not, on their own, have thought to sign up.

A training track for the seniors who orchestrate, next. This is the strategic track of the whole scheme, because it is the one that unlocks the competitive shift described above. It targets partners in professional firms, heads of small and medium-sized businesses, doctors in charge of practices, winegrower-cellarers, independent hoteliers, and experienced staff in municipal and cantonal administrations. Its aim is to teach them not simply to use the tools — many already do — but to architect them within their trade: how to break down a complex problem, how to build a production chain in which the human retains final judgement, how to check the quality of what is produced, how to train their teams to execute within that architecture. This track needs to be longer, denser, better supported, and priced at a level that marks its value — a few thousand francs per participant, funded partly by the organisations they belong to and partly by targeted public schemes. It must also allow seniors from different trades to meet one another, because learning to architect draws as much on cross-disciplinary exchange as on subject-specific content.

Finally, a scheme for vulnerable groups, which chapter 13 will discuss at greater length. Low-skilled or unskilled workers whose jobs will be most exposed — labourers, entry-level administrative staff, low-value-added services — need a dedicated scheme combining AI training with support toward less exposed trades or positions. This track is the hardest to design, because it touches the political question of inequality that the AI transformation will deepen if nothing is done. It must connect the alpine campus with employment services, reskilling bodies, and federal and cantonal social schemes. Its public visibility must be strong, because that visibility is the condition for democratic support for the transformations under way.

These three tracks do not address the same audiences, do not draw on the same content, and do not follow the same funding arrangements. But they can be conceived together, under a single institutional banner, because they answer the same question: how to equip Valais's workers for the transformation ahead. It is the coherence of this arrangement that would make the alpine campus something more than one more fragmentary scheme.

Who should run such a scheme

The question of who carries this scheme is more political than technical. Several configurations are possible, and each has its merits.

A dedicated cantonal initiative, carried by a new structure created for the purpose, would have the advantage of visibility and coherence, but the drawback of cost and delay. Building a new public structure takes two to three years, and it is not obvious it would do better than existing players.

A mandate entrusted to HES-SO Valais-Wallis, through an explicit cantonal mandate and a dedicated budget, is another path. HES-SO has the institutional legitimacy, the pedagogical infrastructure, the links with higher-education institutions and professional associations, and it is already engaged in continuing education. Given a clear mandate and additional resources, it could carry the alpine campus as I describe it. This is the most realistic option, on condition that the mandate explicitly takes on the orchestrating dimension and does not shrink into a mere extension of the existing offer.

A cantonal cooperation among several players — HES-SO, Idiap, professional associations, public operators, authorities — formalised through an agreement that distributes roles and funding, is a third path. Lighter than a new structure, more distributed than a single mandate, but harder to govern. This option can complement the previous one: HES-SO as the driving force, the other players as contracted partners.

Leaving it to the market, supported indirectly through training-financing instruments — training vouchers, subsidies to learners, tax relief — is the last possible configuration. It is the most liberal option, but it does not solve the problem of legibility (learners still would not know where to turn), and it leaves content production to providers who do not all have the necessary expertise.

For my part, I lean toward a combination of the two middle options: a primary mandate to HES-SO Valais-Wallis, coupled with specific agreements for audiences and content it cannot cover alone. This configuration is pragmatic, achievable within eighteen to twenty-four months, and compatible with cantonal resources.

The particular role of the Idiap

I want to dwell for a moment on the role the Idiap could play in this architecture, because it is poorly known outside the canton and under-used within it. The Idiap is one of Switzerland's two or three most active independent AI institutes. Affiliated with EPFL but autonomous, based in Martigny, employing around a hundred researchers and engineers, it has worked for thirty years on subjects now at the heart of the transformation: speech recognition, natural language processing, computer vision, generative models. Its presence in Valais is an opportunity the canton has not fully seized.

Within the scheme I am describing, the Idiap has at least three potential roles. It can produce continuing-education modules for skilled workers — managers, engineers, advanced professionals — who need a solid scientific grounding rather than mere awareness-raising. It can advise the canton on structural technical choices — which tools, which infrastructure, which partnerships — bringing expertise independent of commercial vendors. And it can contribute to the targeted projects mentioned in the previous chapter — regional-language models, applications specific to Valais trades — which require research and development that engineering schools alone do not cover.

Bringing HES-SO and the Idiap together around a shared mission of training and applied innovation for the Valais transformation would, to my mind, be one of the shrewdest moves the canton could make over the next ten years. It does not call for disproportionate resources. It calls for a political decision and an agreement.

A word on compulsory schooling

I will not address in this chapter the question of compulsory and secondary education — primary school, the orientation cycle, upper secondary colleges, vocational schools. It follows its own logic, federal and cantonal curricula, subject didactics, school governance, all of which lie beyond the scope of this essay. Two points are worth noting nonetheless.

Young people in Valais entering primary school today will graduate from secondary education around 2040. On that horizon, predicting precisely which trades and which tools will dominate their working lives is close to impossible. That does not mean we should give up on preparing them; it means we must prepare them to learn throughout their lives, to reason with tools whose form they do not yet know, to exercise critical judgement toward automated output of varying quality. These skills are less technical than one might think. They are, for the most part, what good education has always aimed at.

The debate over AI in schools — ban it, allow it, frame it — is probably poorly posed as long as it fails to distinguish technical learning, which consists of learning to use these tools and which will be necessary, from cognitive learning, which consists of learning to think without them and which will remain equally necessary. Both are indispensable, and they are not won in the same lessons. The canton, through HEP-VS and the cantonal education services, will have to settle these questions in the years ahead. That is not the subject of this essay, but the question exists, and it is more complex than it appears in the shallow public debate.

Why now

The AI transformation is not waiting for the canton to get its house in order. It is unfolding within businesses, within administrations, within professional firms, at the pace at which the tools improve — that is, fast, faster than any previous technological transformation. Workers who learn to use them now gain a lead that will be hard for latecomers to close. And territories that equip their workers now capture a productive momentum that will show up, ten years from now, as highly visible gaps in regional prosperity.

Valais does not need to compete with Zurich and Lausanne on cutting-edge scientific AI research; it would have neither the means nor, as we have seen, the strategic interest to do so. But it needs, and has the means, to equip its working population quickly and broadly. That equipping is not a matter of heavy infrastructure, but of training, orchestration, accessibility. The instruments exist. What is missing is the decision that ties them together into a programme.

And what is missing, above all, is the insight that priority must go to the seniors who orchestrate, rather than to the young who execute. That is the reversal of perspective that will have to be owned publicly for the programme to reach its full effectiveness. A continuing-training policy presenting itself as aimed at experienced seniors has little precedent in Switzerland. It will not be spontaneously popular. It may look counter-intuitive to elected officials accustomed to other priorities. But it is the condition for the canton to capture the competitive shift now opening before it, for this technological transformation to be the occasion for a leap forward rather than one more cause of quiet erosion.

Launched in 2026 or 2027, this programme can produce visible effects as early as 2028, and become structurally significant by 2030. That is a short window, but it is open. One would do well to be among the first to step through it. Valais has shown, in its history, that it knows how to do this when it decides to: with hydroelectric power at the start of the twentieth century, with quality winemaking in the 1990s, more recently with the reorganisation of cantonal hospital services. The moment has come to do it again, on what I believe will be the most structurally important economic transformation of the early part of this century.