No first job, no career: How AI is changing the way careers begin

A third fewer junior jobs: The figure has been making the rounds in headlines and industry association papers since the summer. It comes from the AI report published by the job platform jobs.ch and measures something very specific. In 2025, the share of postings for entry-level candidates among all job postings was 32% lower than the average for the years 2019 to 2022. That is a warning sign. It is not proof that artificial intelligence is taking jobs away from young people.

Anyone debating AI and entry-level jobs in Switzerland can now draw on a surprising amount of data. The data paint a more nuanced picture than the headlines. Unemployment in occupations whose tasks can largely be performed by generative AI has risen considerably faster than elsewhere. University graduates in technology and IT and in economics are having a harder time finding a job. Registered youth unemployment, by contrast, is moving in lockstep with the rest of the labor market, although in the federal government’s surveys, joblessness among young people rose considerably faster at times in 2025. And apprenticeships, in which young people learn an occupation at a company while attending vocational school as part of Switzerland’s system of vocational education and training (VET), show no sign of an AI dent so far. According to the State Secretariat for Education, Research and Innovation (SERI), about two-thirds of young people choose this path.

The figure everyone cites

The jobs.ch AI report analyzes 7.3 million job postings that appeared on jobs.ch, jobup.ch and JobScout24.ch between 2019 and 2025. It compares a pre-AI period, 2019 to 2022, with the period that followed, 2023 to 2025. The finding on entry-level positions covers all occupations. It measures the share of postings, not their number. So if fewer positions are advertised overall and companies cut back above all on entry-level hiring, that share falls. In the public debate, for instance at the business federation Economiesuisse, this became a decline in the number of job postings “since the end of 2022.” That is not what the report says.

Under the jobs.ch definition, roles in IT, administration and human resources, banking and finance, marketing, purchasing and sales count as AI-exposed. They account for 44% of all postings. According to the report, the share of entry-level positions has fallen relatively more there than in the rest of the market. Marco Bertoli, head of JobCloud, the company behind jobs.ch, sums it up: “Anyone starting out in working life today faces a different labor market than just a few years ago.”

Then there is the mood. According to the report, 41% of respondents under 25 have great or very great concerns about being partly replaced by AI in the future. Among 25- to 34-year-olds, the figure is 37%. The survey was conducted via JobCloud’s platforms and newsletters and, by the company’s own account, is not representative of the population. It is still useful as a snapshot of sentiment.

What the unemployment statistics show

The toughest test for the theory that young people are being pushed out is the official data. In September 2022, about two months before the launch of ChatGPT, the State Secretariat for Economic Affairs (SECO) counted 9,368 unemployed 15- to 24-year-olds. In September 2026, there were 14,346. That is an increase of 53%. Over the same period, the total number of registered unemployed rose from 89,526 to 140,480, or by 57%. The youth unemployment rate climbed from 2.1% to 3.3%, the overall rate from 1.9% to 3.0%.

So there are considerably more unemployed young people than four years ago, but among the registered unemployed, their numbers have risen no more steeply than everyone else’s. Their share of the unemployed was 10.5% in 2022 and 10.2% in 2026. A wave of AI that specifically hits the youngest entrants would have to push this share up. That has not happened so far. This figure does not, however, capture university graduates older than 24.

Yet the unemployment statistics have blind spots. They only count people registered with a regional employment center (RAV). Anyone looking for a job after completing their studies without having worked before receives at most 90 daily allowances, and only after a waiting period of 120 days. Many jobless people are not registered with an RAV at all, as the considerably higher unemployment rate calculated by the Federal Statistical Office (FSO) shows. This rate, based on the definition of the International Labour Organization (ILO), is derived from surveys.

The ILO unemployment rate for 15- to 24-year-olds rose from 5.9% to 7.6% between the second quarter of 2024 and the second quarter of 2025, considerably more than the overall rate, which increased from 4.0% to 4.6%. In the fourth quarter of 2025, it reached 8.9%, up from 8.3% a year earlier. That was the same increase in percentage points as for the overall rate, which climbed from 4.4% to 5.0%. By this measure, then, young people were hit harder at times. In the second quarter of 2026, the overall rate stood at 4.9%, up from 4.6% a year earlier.

There is also a seasonal pattern. Youth unemployment rises every summer because many young people look for a job after completing their apprenticeships, as SECO explained in previous years. In 2026, the number of registered unemployed under 25 rose by 2,384 from July to August, an increase of 19.4%. Compared with August 2025, the increase was 7.3%, against 7.1% for all age groups.

KOF finds traces

One of the most important Swiss studies comes from Jeremias Kläui and Michael Siegenthaler of the KOF Swiss Economic Institute at ETH Zurich. They compared occupations whose tasks can largely be performed by generative AI with occupations that are little affected. The result: Since fall 2022, the number of unemployed job seekers in highly exposed occupations has risen up to 27% more steeply. Among those under 50, it has more than doubled in these occupations. In 2024 and 2025, job postings in heavily affected occupational fields reached only 60% to 70% of their initial level. “Overall, however, the results indicate quite strongly that generative AI is behind the recently observed rise in unemployment in some highly AI-exposed occupations,” Siegenthaler said when the study was published.

Siegenthaler updated the analysis in 2026. In a guest article for the IT trade publication Netzwoche, he now puts the figure at up to 30%. According to Siegenthaler, the number of unemployed software developers nearly tripled from the end of 2022 to the end of 2025. “There are hardly any plausible alternative factors that could explain this abrupt increase,” he writes.

The authors nevertheless remain cautious. Their analysis, they note, is not conclusive proof of the influence of AI. It measures shifts between occupations, not the overall effect: The number of people employed in Switzerland rose by 7.4% between the third quarter of 2020 and the second quarter of 2025. And one detail matters for the question of entry-level workers. The study excludes 15- to 19-year-olds because the number of cases is too small. It contains no separate analysis of apprenticeship graduates.

The canton of Zurich’s economic monitoring report (Zürcher Wirtschaftsmonitoring) of September 2026 shows a similar pattern. In the canton, the number of job seekers in AI-exposed occupations rose by 118% from October 2022 to June 2026, compared with 13% in occupations with low exposure. The drivers were software development, general office work and marketing. No causal link can be inferred from this, the canton stresses.

In its country report of September 2026, the International Monetary Fund estimated how widespread the exposure is. According to the report, about 68% of Swiss workers are employed in occupations with significant AI exposure. About a quarter are highly exposed, while AI complements their work only to a limited extent. Such jobs in finance and insurance, in information and communication and in hospitality alone account for about 13% of total employment, the report says. It does not examine entry-level positions separately.

Where it gets tight: in the office and at universities

The clearest warning signs concern not apprentices but university graduates. According to the FSO’s graduate survey of Aug. 27, 2026, 6.4% of bachelor’s graduates of universities of applied sciences in technology and IT were unemployed one year after graduating; for the class of 2022, the figure was 3.0%. Among university master’s graduates in economics, the rate rose from 3.5% to 8.5%, and across all university master’s degrees from 3.9% to 6.4%. Respondents most often cited a lack of professional experience as the reason for their difficulties, along with the job situation in their field and the state of the economy. AI does not appear in the report as a cause.

For its Swiss AI Jobs Report 2026, the Lucerne University of Applied Sciences and Arts (HSLU) analyzed job postings for AI specialists. From July 2025 to June 2026, just under 5,000 such positions were advertised, 32% more than in the previous year, according to the study. Demand for entry-level candidates also increased, but only 5% of the postings were permanent junior positions. Positions for experienced specialists rose by 42%. Even for entry-level positions, two to three years of professional experience are often required, the study says. “People starting their careers describe the job market as extremely competitive,” says study author Curdin Derungs.

Economist Mathias Binswanger of the University of Applied Sciences and Arts Northwestern Switzerland (FHNW) sees newcomers in particular under pressure, because in management they are often entrusted with research, reports and market analyses. “It is precisely these activities that can easily be done with AI, which makes it harder for newcomers,” he told the news site Nau.ch in April. Consulting firms sound a similar note. According to Adrian Jones, a partner at PwC Switzerland, AI is changing how careers begin and how expertise develops. And entrepreneur Christian Bartholet, head of Grass & Partner AG, states: “In my assessment, the situation is currently particularly challenging for people starting their careers.”

Apprenticeships as a buffer

The picture is entirely different for apprenticeships, which about two-thirds of young people in Switzerland choose, according to SERI. The transition barometer (Nahtstellenbarometer), for which the research institute gfs.bern surveyed about 7,500 young people and 5,000 companies on behalf of SERI from February to April 2026, reports a stable apprenticeship market. At the time of the survey, companies were offering about 74,000 apprenticeship positions, according to extrapolated figures. Among the companies, 72% advertised the same number of positions as in the previous year, while 10% each advertised more or fewer. According to the barometer, commercial training, the office apprenticeship known in Switzerland as KV, remains in particularly high demand.

By the end of May, more than 54,500 apprenticeship contracts had been signed, about as many as a year earlier. By the end of August, about 76,000 apprenticeship contracts had been concluded nationwide, in line with previous years, according to SERI. The August survey of the transition barometer, whose results SERI published on Oct. 29, 2026, shows the same picture. Companies kept their offering steady: About 86,000 apprenticeship positions were on offer, and by August more than 75,000 had been filled, which corresponds to a fill rate of 88%. Of the more than 91,000 young people who finished compulsory schooling in summer 2026, 49% chose initial vocational education and training, 34% a general education path and 17% an interim solution; according to the public broadcaster SRF, the shares were unchanged from the previous year. AI is not mentioned in the federal government’s press releases on apprenticeships from June and September, nor in the report on the August survey.

Experts therefore do not yet see the pressure reaching the apprenticeship market. “The supply in the KV field does not appear to be declining, or not yet significantly,” Urs Casty, founder of the apprenticeship platform Yousty, told the business magazine Bilanz in August. Education economist Stefan C. Wolter of the University of Bern points to the companies: “If something happens on a larger scale in the apprenticeship market, it has to come from the companies.” Patrick Leisibach of the think tank Avenir Suisse states: “At present, the pressure of a profession being at risk has not yet arrived in many areas.” Alexander Bélaz, president of the employees’ association Angestellte Schweiz, puts it in fundamental terms in a guest article: “AI rarely replaces an entire profession, but rather individual tasks, routines, learned patterns.”

Still, there is a first warning sign. In August 2026, it emerged that BICT, a training company in Bern, would have to close. About 120 young people training in IT and mediamatics feared for their apprenticeships. Besides a COVID-19 loan, the economic situation and cost-cutting at Swiss Post and the federal government, the founder also cited AI as one of several reasons. According to a Netzwoche update of Sept. 11, 2026, the Technische Fachschule Bern, a technical training school, will take over the 80 apprentices in their first and second years by November 2026.

Comparisons with the U.S. studies therefore have their limits. The much-cited Stanford data concern 22- to 25-year-olds in the U.S. In Switzerland, about two-thirds of young people opt for initial vocational education and training, according to SERI. There is much to suggest that a company that trains apprentices does the math differently: It also trains them in order to develop its own skilled workers. That calculation is likely to change more slowly with AI than that of a company expected to hire a fully trained junior employee.

AI and entry-level jobs in the U.S.: What the Stanford data show

One of the most widely cited international studies is titled “Canaries in the Coal Mine?” and comes from Erik Brynjolfsson, Bharat Chandar and Ruyu Chen of the Stanford Digital Economy Lab. The researchers analyze wage data from the payroll processor ADP covering millions of workers in the U.S. In the August 2025 version, they estimated a relative decline in employment of 13% for 22- to 25-year-olds in the most exposed occupations. In the revision of Aug. 12, 2026, they use a simpler measure. By that measure, employment in this group is 19% below the level it would have reached had it kept pace with low-exposure occupations. With data through July 2025, the figure on this measure was 15%.

According to the data, the gap stems mainly from fewer new hires of young workers, not from more departures. The authors do not, however, see broad displacement. They stress explicitly that their work does not estimate a causal effect of AI. An analysis of more than 21,000 U.S. companies by the corporate card provider Ramp and by Revelio Labs, which Economiesuisse cited in July, reaches a different conclusion. Companies that use AI intensively increased employment by about 10% in the two years after adopting it, and entry-level positions by about 12%.

Another study puts the concern about graduates into perspective. According to a discussion paper by Robert Fairlie and Jane Wu (University of California, Los Angeles), published in September 2026 in the discussion paper series of the economists’ network IZA@LISER, unemployment among 22- to 25-year-old college graduates with a bachelor’s degree in the U.S. did not jump in summer 2026 compared with previous summers. No significant link to the AI exposure of occupations emerges, and these authors, too, draw no causal conclusion.

The economy, tariffs, India: the other suspects

The timing alone suggests that AI is unlikely to be solely responsible. The rise in unemployment coincided with a period of weak industrial activity and, from 2025, of the tariff dispute with the U.S. KOF found a negative effect on job postings as early as the pandemic year 2020, that is, before ChatGPT. Then there is offshoring. According to Nau.ch, UBS is cutting thousands of jobs in Switzerland and plans 3,000 positions in India, but stresses that the two have nothing to do with each other. Swisscom is setting up a site in Lisbon for services in IT, software development and finance, with about 40 employees from 2027 and about 200 in the medium term, according to the financial news agency AWP. A Nau.ch report says offshoring to India particularly affects administrative and IT-related functions, including low-level positions.

The telecom company Sunrise shows how hard it is to disentangle the causes. On Oct. 23, 2026, it announced that it would lay off 358 employees for operational reasons. In September, there had still been talk of up to 450 full-time positions. According to Sunrise, the goals of the restructuring include substantial cost savings. ICTjournal, a trade portal in French-speaking Switzerland, reports that the company counts AI and data-driven automation among the tools (“outils”) it intends to use in the restructuring. When it made its announcement in September, Sunrise fully exempted apprentices from the job cuts.

On the other side of the equation is demography. According to Deloitte, about 900,000 workers will reach retirement age in the next 10 years, about one-fifth of the workforce. In nursing and care, the skilled trades, transportation and education, the share of workers over 55 is particularly high, according to the study, while the automation potential of today’s AI is low. Deloitte expects a growing labor shortage there despite AI. Switzerland’s competitiveness depends on how quickly and how effectively companies integrate generative AI into their daily work, says Deloitte chief economist Michael Grampp: “They need to use it to become more productive and to train their employees for changed tasks.” His colleague Marc Beierschoder, head of AI and data at Deloitte Switzerland, warns of a different mistake: “The greatest danger is not that AI replaces jobs.” More dangerous, he says, is that companies leave their ways of working unchanged and simply lay AI on top.

Arena: Is AI taking entry-level jobs away from young people?

What young people and companies can do now

One finding from the field supports the optimists. Amosa, the labor market observatory of several cantons in German-speaking Switzerland, surveyed just under 6,000 job seekers and about 700 companies. Those who use AI regularly received about 48% more invitations to job interviews, according to Amosa. The companies estimated that about 39% of applications had been written at least partly with AI.

Vocational education is responding, too. The association ICT-Berufsbildung Schweiz has created a federal professional certificate, “AI Business Specialist”; according to information from October 2025, the first examination was scheduled for fall 2026. The International Monetary Fund sees Switzerland as well positioned: Its education and vocational training system, the IMF says, is a strong foundation for the use of AI. Integrating digital and AI-related skills earlier could prepare Switzerland even better, it writes. Above all, the fund adds, affordable continuing education in mid-career would help workers adapt to changing requirements.

Commentary: Don’t dismantle the ladder

AI is not robbing young people in Switzerland of their future. Those who claim it is must explain why registered youth unemployment has not risen faster than overall unemployment since 2022 and why companies are offering as many apprenticeship positions in 2026 as in the previous year. But those who dismiss the warning signs must explain why more and more IT specialists and economics graduates are out of work, why joblessness among young people rose faster at times in the federal government’s surveys and why only 5% of postings for AI specialists are permanent junior positions.

The truth lies in the office. There, the machine is taking over the routine work through which generations of newcomers worked their way up: minutes, research, first drafts, simple programming. A company that leaves these tasks to AI and therefore stops hiring juniors saves money in the short term. In the long term, it is sawing off the branch it sits on. Experienced professionals do not fall from the sky. They emerge from newcomers whom someone gave a chance.

With its apprenticeship system, Switzerland has a model that anchors precisely this way of thinking in companies. It would be a mistake to cut back on it now, of all times. The responsibility lies with the companies that train apprentices and with young people themselves. They must learn to master the new tools instead of fearing them. The state does not need to issue new regulations for this. What it must do is ensure that curricula and continuing education keep pace with technology.

All figures as of Oct. 30, 2026. Unemployment figures from SECO (registered unemployed, most recent published month September 2026), ILO unemployment rates from the FSO (definition of the International Labour Organization). The ILO unemployment rate for 15- to 24-year-olds is given through the fourth quarter of 2025, the most recent figure we were able to verify in an FSO press release. The press releases on the first and second quarters of 2026 give no figure for this age group. The results of the August survey of the transition barometer are taken from SRF’s report of Oct. 29, 2026, on the SERI press release; we were unable to view the press release itself. The number of unemployed 15- to 24-year-olds in September 2025 is derived from SECO’s year-on-year comparison. Quotes originally given in another language were translated by the editors. Sources are cited in their original language.

Related coverage: The stars who don’t exist: How AI singers are hijacking the charts · Brussels regulates, Zurich builds: How Switzerland is winning the AI race by not taking part (in German) · Social media ban for children: Europe draws the line, Switzerland hesitates

Sources

• jobs.ch: KI-Report 2026, methodology and results (June 24, 2026)
• Netzwoche: KI drängt Nachwuchskräfte aus Bürojobs (June 24, 2026)
• Economiesuisse: KI vernichtet Jobs? Ein Mythos wird brüchig, opinion piece by David Stauffacher (July 16, 2026)
• Netzwoche: Künstliche Intelligenz hinterlässt deutliche Spuren auf dem Schweizer Arbeitsmarkt, KOF (Oct. 27, 2025)
• KOF ETH Zurich: Jeremias Kläui and Michael Siegenthaler, KOF Studien No. 186 (October 2025)
• Netzwoche: guest article by Michael Siegenthaler on rising unemployment (July 22, 2026)
• SECO: Die Lage auf dem Arbeitsmarkt im September 2026 (Oct. 6, 2026)
• SECO: Die Lage auf dem Arbeitsmarkt im August 2026 (Sept. 7, 2026)
• SECO: labor market data, September 2022 (Oct. 7, 2022)
• SECO: labor market data, September 2023 (Oct. 6, 2023)
• Federal SME Portal: Arbeitslosigkeit im September 2024 gestiegen (Oct. 16, 2024)
• Nau.ch: Schweizer Arbeitsmarkt zeigt sich trotz Zollgewitter robust (Oct. 6, 2025)
• FSO: Arbeitskräfteerhebung im 2. Quartal 2025 (Aug. 19, 2025)
• FSO: Arbeitskräfteerhebung im 4. Quartal 2025 (Feb. 17, 2026)
• FSO: Erwerbslosenquote (ILO) im 1. Quartal 2026 auf 5,2 Prozent (May 18, 2026)
• FSO: Erwerbslosenquote (ILO) im 2. Quartal 2026 auf 4,9 Prozent (Aug. 18, 2026)
• Canton of Fribourg: Arbeitslos nach der Ausbildung, entitlement to daily allowances
• Moneycab: SECO on the seasonal rise in youth unemployment in August (Sept. 9, 2020)
• Netzwoche: Kanton Zürich rechnet mit deutlichem Produktivitätswachstum durch KI (Sept. 24, 2026)
• IMF: Switzerland, Selected Issues, Country Report No. 26/234 (September 2026)
• FSO: Erwerbslosigkeit der Hochschulabsolventinnen und -absolventen gestiegen (Aug. 27, 2026)
• Netzwoche: Erwerbslosenquote unter IT-Studienabsolventen verdoppelt sich (Aug. 31, 2026)
• AWP via Cash: Schweizer KI-Stellenmarkt wächst rasant, HSLU (Oct. 8, 2026)
• Netzwoche: Schweizer KMU buhlen um erfahrene KI-Fachkräfte, HSLU (Oct. 9, 2026)
• Nau.ch: Kündigungswelle, wie schlimm steht es um den Schweizer Arbeitsmarkt? (April 19, 2026)
• PwC Switzerland: AI Jobs Barometer 2026 (June 15, 2026)
• Nau.ch: KI im Arbeitsmarkt, besonders für Berufseinsteiger herausfordernd (Sept. 26, 2026)
• SERI: Berufsbildung in der Schweiz 2025
• SERI: Nahtstellenbarometer 2026, berufliche Grundbildung bleibt erste Wahl (June 11, 2026)
• SERI: Lehrstellenmarkt Ende Mai 2026 (June 23, 2026)
• SERI: Lehrstellenmarkt 2026, stabil und vielfältig (Sept. 17, 2026)
• Bilanz: KV-Lehre 2026 sicher trotz KI? (Aug. 11, 2026)
• Netzwoche: Wie KI unsere Arbeit wirklich verändert, guest article by Alexander Bélaz (Aug. 1, 2026)
• Netzwoche: 120 Jugendliche bangen um ihre ICT-Lehrstelle (Aug. 18, 2026, updated Sept. 11, 2026)
• Stanford Digital Economy Lab: Canaries in the Coal Mine?, revision (Aug. 12, 2026)
• Stanford Digital Economy Lab: revised paper, PDF (Aug. 12, 2026)
• IZA: Robert Fairlie and Jane Wu, The Early Impacts of AI on Employment among Recent College Graduates, Discussion Paper No. 18945 (September 2026)
• Nau.ch: Immer mehr Schweizer Firmen lagern Jobs nach Indien aus (March 23, 2026)
• AWP via Cash: Swisscom plant neuen Standort in Lissabon (Aug. 6, 2026)
• Netzwoche: Sunrise baut definitiv Stellen ab, jedoch weniger als zunächst geplant (Sept. 21, 2026, updated Oct. 13 and Oct. 26, 2026)
• ICTjournal: Sunrise précise la place de l’IA dans son plan de restructuration (Oct. 12, 2026, in French)
• SRF: 49 Prozent machen Lehre, Schweizer Unternehmen halten ihr Lehrstellenangebot konstant (Oct. 29, 2026)
• Netzwoche: Generative KI kann die Pensionierungswelle nicht abfedern, Deloitte (Oct. 22, 2026)
• Netzwoche: KI öffnet Stellensuchenden mehr Türen zu Bewerbungsgesprächen, Amosa (April 15, 2026)
• Netzwoche: Fachausweis “AI Business Specialist” startet 2026 (Oct. 14, 2025)