The skills that will make the difference in the AI era — and why Switzerland is not immune

Artificial intelligence doesn't take whole jobs. It takes tasks. The distinction matters, and it changes how we should think about employment, training and competitiveness — in Switzerland just as much as in Valais.

AI's impact on Swiss employment: the wave has already begun

According to AXA's annual study of the labour market (November 2025), the share of Swiss SMEs that have integrated AI rose from 22% to 34% in one year. And 57% of employers observe efficiency gains thanks to these tools, against 46% the previous year. Adoption is accelerating.

According to the OECD Skills Outlook 2025, technologies such as large language models will modify the skill profiles required for around 40% of jobs, partly through automation, partly through the emergence of new tasks. That isn't 40% in job losses; it is 40% in transformation. Positions don't disappear overnight, but their content changes faster than the people who hold them.

On the Swiss side, Avenir Suisse's figures are unambiguous: 80% of office employees are potentially exposed to AI competition, around 490,000 people nationwide, including 380,000 administrative employees without specialisation. At the same time, PwC Switzerland's AI Jobs Barometer 2025 shows that AI-related job offers have multiplied by ten since 2018, going from 2,000 to 20,000 in 2024. AI destroys on one side and creates on the other, but rarely for the same profiles.

In Switzerland, one might be tempted to think that our economic fabric (solid, diversified, high value-added) protects us. That is partly true. But the mechanism at work does not target sectors: it targets tasks within each profession. A fiduciary firm in Sion, an architectural office in Martigny, a property management firm in Verbier, a wine estate in Fully: all have administrative, analytical or editorial layers in their daily operations that AI can already handle.

The real risk is not abrupt replacement. It is gradual erosion: one fewer employee here, one outsourced mandate fewer there, a junior position not refilled elsewhere. In Valais, the unemployment rate rose from 2.9% in January 2025 to 3.3% in December, with a 14.1% increase in jobseekers in the fourth quarter compared with the previous year (CCI Valais / BAK Economics). It is hard to isolate AI's share in this trend, but the signals converge.

Early-career profiles are the first affected, not by layoffs, but by silent non-employment. This phenomenon also affects the developer profession, where entry-level tasks are precisely those AI automates best. The tasks that historically allowed people to prove themselves are precisely those AI handles best. And when you are told "come back with experience", the question remains: how do you acquire it without a first job?

Three traps to avoid when thinking about AI and employment

"AI will create as many jobs as it destroys"

This is the most frequent argument, and the most misleading. Take an example every Valais resident understands.

The Valais vineyard covers 5,000 hectares, 22,000 owners and more than 80,000 plots spread across 65 wine-growing municipalities: a fabric of micro-holdings passed down across generations, on often steep slopes supported by 3,000 kilometres of dry stone walls. Today, autonomous viticultural robots mechanically weed the rows without a driver. Drones equipped with multispectral cameras map the health of vines down to the centimetre. Predictive algorithms determine the optimal harvest date by combining weather data, harvest history and ripening curves, with a precision no human eye can match across 80,000 plots.

The winegrower who, five years ago, hired three seasonal workers for leaf-thinning and weeding can now cover part of these tasks with an autonomous robot. These three positions have not been "replaced" by three positions of viticultural-robot programmers. They have simply disappeared. A productivity tool, by definition, lets one person do the work of several.

What is true in the vineyards of Fully or Chamoson is just as true in the offices of Sion or Martigny. Those who will build tomorrow's AI models will be few, extremely qualified, and the seats will be scarce.

"Soft skills will be enough"

The idea is not wrong, but it is incomplete. Relational skills, listening, the ability to understand a client's need have real and growing value. But making them the sole answer to the upheaval underway is treating the symptom without addressing the cause. Soft skills are necessary, but not sufficient. In writing, for example, AI and writers are becoming strategic allies rather than rivals, provided you master both dimensions. The market increasingly values hybrid profiles: according to KOF (ETH Zurich), people able to combine domain expertise with mastery of AI tools are the most sought-after and the rarest in Switzerland in 2025.

"I'll wait and see"

This is the most dangerous trap. The paradox is already measurable: 65% of Swiss companies have integrated AI into their long-term strategy, but only 13% operate with clearly defined performance indicators (Z Digital Agency, 2025). AI enters organisations faster than organisations adapt to it. And employees who wait for their job description to evolve before starting to develop new skills take a major risk.

When an executive sees that AI can absorb part of the team's tasks, they do not redistribute the time saved into personal development. They resize the team.

The six skills AI cannot replace

The real challenge is not to compete with the machine on its turf. It is to develop what it does not master. Six skills stand out, all deeply human.

1. Critical thinking

Not systematic doubt or surface scepticism, but the ability to question one's own biases, to change one's mind in the face of new information, to question a statement rather than execute it blindly. In Switzerland, our culture of consensus and rigour is an asset, provided it does not become rigidity.

A language model never does any of that. It generates the statistically most probable answer. It does not doubt.

2. Learning to learn

It has never been so easy to acquire a new skill independently. Yet our education system (in Switzerland as elsewhere) remains largely structured around the recall of fixed knowledge. The decisive skill is no longer mastery of a given field, but the demonstrated ability to master a new one quickly.

Swiss employers have understood this: according to PwC's AI Jobs Barometer 2025, practical experience in AI is now more valued than university qualifications in AI-related job offers.

3. Human-machine collaboration

Using AI is not just copy-pasting prompts found online. It is understanding how a language model works, knowing its limits, knowing how to phrase a request to obtain a usable result, and keeping your judgement before and after every interaction.

In Switzerland, 34% of SMEs already use AI to automate certain work steps, against 23% a year earlier (AXA study, 2025). But only one third of companies have set clear rules on the data employees can input into these tools.

4. Emotional intelligence

Sensing, intuiting, perceiving what is not said, detecting an inconsistency through "feeling" before being able to articulate it: all this remains beyond AI's reach. In a canton like Valais, where the economic fabric rests on more than 31,000 establishments, the vast majority of which are micro-enterprises with fewer than 10 people, close relationships are the norm.

The provider who understands their client beyond the brief, who senses when something is off, who adapts their approach in real time: that one has a future, whatever their profession.

5. Divergent creativity

Not the ability to produce twelve variations of the same visual: AI already does that very well. Real creativity is the kind that creates unexpected links between distant fields, that formulates questions no one is asking, that proposes approaches existing data does not suggest.

AI is trained on what already exists. It excels at interpolation, i.e. producing something coherent between two known points. But extrapolation, which consists of imagining a point no one has yet mapped, remains a human advantage.

6. Ethics and understanding AI's limits

Knowing where AI is reliable and where it is not. Understanding data privacy issues, algorithmic biases, regulatory compliance, particularly in the Swiss context with the nFADP (new Federal Act on Data Protection, in force since September 2023). Only one third of Swiss SMEs have set clear rules on AI use by their employees. This skill must not serve as a pretext for delaying adoption. Compliance can be learned; inaction in the face of transformation, by contrast, has a price.

What this concretely means for Valais SMEs

Valais has specific assets in this transition. Its economy rests on 31,356 establishments, 89% of which are SMEs with fewer than 250 employees. More than 27,000 micro-enterprises employ a third of the canton's working population.

Entire sectors (tourism, with 4.4 million overnight stays in 2024, viticulture, with 5,000 hectares constituting Switzerland's largest vineyard, construction, real estate) rest on human relationships and field judgement. AI does not reproduce these skills.

But these assets only protect those who actively seize them. The fiduciary firm that does not train its employees on AI will see its margins erode against more productive competitors. The real-estate agency that does not integrate automation into its administrative processes will lose competitiveness. The engineering office that does not invest in human-machine collaboration will fall behind on deliverables.

The signal is clear: in Valais, real GDP growth reached only 0.8% in 2025, against 1.4% for Switzerland as a whole. The canton cannot afford a further delay in adapting its skills. BAK Economics forecasts a rebound to 1.3% in 2026, provided Valais businesses invest in productivity and innovation.

The challenge is not technological. It is human, plain and simple. It is about developing, now, the skills that will make it possible to adapt to a permanently changing environment.

A new CV for a new world of work

Tomorrow's CV will no longer list only qualifications and years of experience. It will have to demonstrate a capacity for adaptation, evidence of self-learning, an aptitude for collaborating with AI tools, and measurable relational intelligence.

What counts is not what you can do today. It is your demonstrated ability to learn what you do not yet know.

FAQ

What skills should I develop in the face of AI in Switzerland?

Six skills make the difference: critical thinking, the capacity for self-learning, human-machine collaboration, emotional intelligence, divergent creativity, and an understanding of AI's ethical and regulatory limits (nFADP in Switzerland, GDPR for European clients). These skills are complementary to technical mastery, they do not replace it.

Will AI eliminate jobs in Valais?

AI does not eliminate whole professions; it transforms the tasks that make them up. In Switzerland, Avenir Suisse estimates that 490,000 office employees are potentially exposed. In Valais, where 89% of jobs are in SMEs, the impact will take the form of gradual erosion rather than mass redundancies: junior positions not refilled, administrative tasks absorbed, mandates outsourced less often. Sectors where the human relationship is central (tourism, viticulture, real estate) will retain an advantage, provided they combine relational skills with mastery of digital tools.

How can Valais SMEs prepare for AI?

Three priorities. First, audit your digital visibility: how does your business appear when a potential client asks ChatGPT or Google a question? It is measurable: the GEO Score™ assesses a company's citability by AI on a 0 to 100 scale. Second, train your teams on human-machine collaboration, not a theoretical seminar, but a guided daily practice. Third, identify short-term automatable tasks (administrative, editorial, analytical) to free human time for high value-added work.

MCVA Consulting SA supports Swiss companies in their AI transformation, from citability audits (GEO Score™) to upskilling teams. Contact us for a no-obligation conversation.