"The Machines Are Coming": AI and the Labour Market

One Monday Morning

Picture Anna, a 32-year-old marketing professional in Budapest. In 2023, she worked alongside four colleagues at the Hungarian subsidiary of a multinational, writing newsletters, producing social-media posts and analysing campaigns. Today, in May 2026, she sits alone in the office. Her four colleagues were let go last autumn — not because they were bad at their jobs, but because a corporate AI tool, its monthly licence fee a fraction of one employee's salary, now writes, analyses and sends out a hundred newsletters a day. Anna merely supervises the process.

Anna still has a job. But she no longer creates. She supervises.

This is not science fiction. This is the Hungarian labour market in 2026. And Anna's story is one of millions.

By the Numbers

According to the International Labour Organization's latest report, 186 million people will remain unemployed worldwide in 2026, and youth unemployment has climbed to 12.4%. That is roughly the 2024 level — on its own, nothing alarming. The real worry is not the headline figure but what lies behind it: the ILO warns that the composition of unemployment is changing, with a growing share of structurally displaced workers — people made redundant not by the business cycle but by technology. Artificial intelligence and automation, the organisation says, are undermining job quality, especially in highly skilled, white-collar professions.

Goldman Sachs's March 2026 analysis paints an even starker picture. According to the investment bank, two-thirds of jobs in the United States and Europe are "exposed to some degree of AI automation", and roughly a quarter of work tasks could be carried out entirely by algorithms. In the long run, 6–7% of American workers — some 11 million people — could be replaced outright by machines. In the same analysis, Goldman also points out that if AI's productivity gains materialise, global GDP could rise by as much as 7% within a decade. The debate is not about whether there will be growth; it is about who loses, and how much, during the transition — and on current trends the burden is not being borne evenly.

By 2025, 51% of companies in McKinsey's global survey reported that generative AI was reducing their need for entry-level positions. In the tech sector, there is no way to dress it up: US tech firms announced 154,445 lay-offs in 2025, up 15% on the previous year. The first quarter of 2026 brought a further 52,050, 40% more than in the same period of 2025.

The biggest tech giants are no exception:

• Microsoft laid off 15,000 people in 2025, then in April 2026 announced a voluntary redundancy programme for 7% of its US workforce
• Meta eliminated 8,000 jobs in April 2026 — 10% of its total headcount
• IBM cut 8,000 administrative positions while recruiting engineers

Notably, the tech sector's first great wave of lay-offs, in 2022–23, cannot be blamed on AI alone. In the years of zero-interest-rate policy (ZIRP), companies gorged on artificially cheap credit and bulked up their headcounts beyond reason — when central banks began raising rates in 2022, it was this surplus they shed, with AI often merely supplying the "efficiency" narrative to dress up the dismissals. The 2025–26 wave is different in kind: AI is no longer a PR cover story but is genuinely embedding itself in organisational structures — and the slump in job postings bears this out.

As we put it in that series: "Keep staff, burn cash. Run algorithms, wear the crown."

America, Europe, Hungary: A Three-Speed World

The United States: The Tech Giants' Playground

In the United States, the process accelerated in the second half of 2025. According to Goldman Sachs's chief economist Jan Hatzius, AI investment had a "basically zero" effect on GDP growth in 2025 — the overwhelming bulk of the money went on hardware bought abroad, mostly from Taiwanese and Korean chipmakers. On the labour market, however, the effect is already being felt: companies that cited AI in justifying their lay-offs cut job postings by 12%, compared with an average of 8%.

Europe: Slow Adoption Behind a Bureaucratic Shield

Analysts argue — a point our series has also made — that the EU AI Act's strict rules and CE-marking requirements are temporarily holding back the spread of robots and autonomous systems. A business owner spends weeks filling in risk-assessment paperwork for a robot. Today, that bureaucratic shield protects workers; by 2030, the same over-regulation will be strangling European industry, turning the continent into "an expensive open-air museum of human labour".

Hungary: The Double Exposure of Factory Floor and Middle Class

In Hungary, the picture is peculiar. According to the KSH, Hungary's Central Statistical Office, 4.637 million people were in work as of November 2025 and the unemployment rate stood at 4.4% — low by EU standards. The employment rate among 15–64-year-olds is 75%, a record.

Yet a 2019 analysis by PwC Hungary suggests this stability is illusory. By PwC's estimate, close to 24% of Hungary's workforce — some 965,000 people — work in manufacturing, above all in the automotive sector. A further roughly 477,000 work in retail and wholesale trade, 424,000 in public administration, 271,000 in transport and warehousing and 268,000 in construction.

These are precisely the sectors PwC finds most exposed to automation:

• By 2025 (the augmentation wave — AI assisting the human): 234,600 jobs at risk, chiefly in trade (108,400), public administration (66,000) and administrative services (30,000)
• By 2030 (the autonomy wave — AI working on its own): 922,000 jobs — 23% of total employment — could be automated
• The most exposed: craft and skilled trades workers (209,000 people at 38% automation risk) and machine operators (168,000 at 26%)

A distinctive feature of the Hungarian labour market is its unusually well-developed SSC sector: shared service centres employing young, multilingual staff in large numbers on what is essentially administrative work. That is precisely the segment of the workforce that generative AI targets first.

The Historical Parallel That No Longer Holds

Every technological revolution trots out the same soothing mantra: "Machines take the old jobs but create new ones." In the Industrial Revolution, rural labourers became factory hands. In the computer revolution, factory hands became office workers.

But this wave is different.

The earlier waves created new jobs because the machines extended human capabilities. The steam engine was stronger than muscle, but it needed an operator. The computer was faster than the abacus, but it needed a programmer.

Generative AI, most analysts argue, does not merely extend human capabilities — in routine tasks, it increasingly replaces them. ChatGPT is not a better word processor; it is a system that generates text, analysis and code on its own, with no human input or only minimal supervision. And the newest AI agents already navigate computer systems on their own, write code and improve themselves.

As we showed in the series through an industry model: at a mid-sized firm of 120 employees, 18 remained. The 102 who were let go were not replaced by 102 new jobs. Among the 18 who stayed, there were barely a handful of technical roles — the majority were management and key oversight. Meanwhile, the former 30-strong IT team shrank to three, and the 25-strong analyst team to one.

This arithmetic promises no renaissance. It promises a transition in which millions try to make a living in an economy where demand for labour is falling drastically — and irreversibly.

Where Hungary Stands

Hungary's labour market is in a paradoxical state. On the one hand, unemployment is at a record low (4.4%); on the other, by PwC's reckoning more than a fifth of the workforce is in sectors that automation could hit hard over the next five to ten years.

The regional differences are stark, too. In the dynamic agglomeration formed by Budapest and Pest county, automation risk is just 30.9%; in the peripheral regions of northern Hungary and southern Transdanubia, it reaches 49.8%. Where schooling and mobility are lower, the danger of being displaced by a machine is higher.

Especially vulnerable are women working in public administration and in trade, and middle-aged men with secondary education in manufacturing and construction. They form the backbone of the Hungarian middle class — and they are the ones the machines are targeting first.

What Remains When the Work Is Gone?

Anna's story is not unique. According to KSH figures, 4.6 million people work in Hungary. By PwC's estimate, over the next decade close to a million of them could find that their jobs are being done by algorithms.

The question is not whether this will happen. The question is what happens to those the machines push out.

If work — for centuries the foundation of livelihood, identity and social standing — disappears, what takes its place? Can the state provide for those whom technology has made "redundant" — and does it want to? And if so, at what price?

In the next part, we go looking for the answer: what is basic income, and what does it promise the millions left jobless?