AI and employment debates often show two deviations: one is the panic that "AI will destroy all jobs," the other is the carelessness that "nothing will change." Both are wrong. Recent data from international organizations paints a third, far more complex picture: jobs are not disappearing, they are moving. And the burden of this movement does not fall on everyone equally.
Global forecasts: what the numbers say
Per the World Economic Forum's Future of Jobs Report 2025 (based on a survey of 1,000+ employers from 55 countries), 92 million jobs will disappear worldwide by 2030, but 170 million new ones will be created — a net increase of 78 million. 22% of current occupations will undergo serious change in this period. From AI and information-processing technologies alone, 9 million jobs are expected to disappear and 11 million to be created.
An important aspect: 41% of employers said they plan to cut headcount because AI automated certain tasks. This is not a theory about the distant future — workforce plans are already being rewritten.
Per the IMF (International Monetary Fund), about 40% of global employment is exposed to AI: in advanced economies this reaches 60%, in low-income countries 26%. The IMF notes: in developed countries about half of affected jobs could benefit from AI — the human is not replaced but augmented. The ILO (International Labour Organization) estimates 25% of global jobs as at risk from generative AI.
McKinsey Global Institute research gives an important warning: 57% of work hours in the US could theoretically be automated with existing technologies — but the institute itself comments:
"This is a technical possibility, not a forecast of job loss."
Indeed, only 14% of surveyed organizations reported that headcount decreased due to AI adoption. Most companies use AI not to replace people but to automate routine small tasks and redirect employees to higher-value activity — strategy, creative problem-solving, and client work.
The "entry-level squeeze": the most vulnerable layer
One of the most discussed conclusions of the Stanford AI Index 2026 report is the "entry-level squeeze." Since 2024, the number of software developers aged 22–25 in the US has fallen nearly 20% — other age groups kept growing. In the same period, top models' scores on the SWE-bench programming test rose in a year from 60% to nearly 100%. The connection is visible: AI is taking on exactly the routine code-writing tasks young specialists do.
Per Stanford Digital Economy Lab data from August 2026, employment of 22–25-year-olds in AI-exposed occupations is 19% below expected levels. Per Challenger, Gray & Christmas calculations, US employers directly linked 116,175 of job cuts announced in 2026 to AI — 22% of all cuts.
The conclusion: experienced specialists are boosting efficiency with AI, while new entrants are losing the "ladder" into the profession. This is the most painful point of the labor market and the most urgent problem precisely for countries with large youth populations.
Which occupations are at risk, which will grow
Research unanimously lists the most exposed fields: administrative support, data processing, customer service (call centers), simple content creation, routine financial analysis, and entry-level programming. The common trait of these fields is task repetitiveness and reliance on clear rules.
Growing fields are different: AI system management, data engineering, cybersecurity, "green" technologies, and roles based on human-AI collaboration. The WEF separately emphasizes: construction workers, delivery workers, care workers, and agricultural occupations will keep growing — because building, road-laying, and patient care stay in the physical world.
Healthcare is one of the most stable fields long-term: per the US Bureau of Labor Statistics forecast, nurse practitioner employment will grow 40.1% in 2024–2034 (about 128.4 thousand new positions), and home care and personal assistance workers will add 739.8 thousand new jobs. The reason is simple: when a decision concerns the human body and life, the answer "the algorithm said so" is not enough for society — a human is expected to be accountable.
The Uzbek context: a young labor market and an outsourcing economy
For Uzbekistan this question is especially urgent because the country is demographically very young: the working-age population exceeds 21 million, the average age is 29. The "entry-level" that AI is mastering fastest is exactly the door through which youth enter the labor market.
The positive side — unprecedented workforce training pace: 300 thousand IT specialists work in the country (64.3 thousand in 2017), university students grew from 297.7 thousand to 1.9 million. The "One Million Uzbek Coders" program covered 1.7 million young people, the "Five Million AI Leaders" program trained 1.3 million people. Uzbekistan ranks 1st in the world in learner activity on Coursera — 6.25% of the workforce actively upskills online.
The dangerous side — the economy's structure. IT Park residents' exports (over $1.1 billion) rely mainly on outsourcing and BPO services, nearly half of exports go to North America. Customer service, data processing, entry-level programming — these are exactly the fields most exposed to AI. The "cheap and skilled workforce" advantage erodes as AI gets cheaper. So moving up the value chain is mandatory — from simple execution to design, management, and complex engineering.
The education system also faces a serious task: 1.9 million students study in universities, but most curricula are still based on memorization and repetition — exactly what AI does best. And digitization of government services creates new demand: civil servants who can work with AI and analyze data are needed. The problem is not only in the private sector — the entire workforce training chain must be reviewed.
At the same time there is opportunity: selling AI automation services to foreign clients is itself a new market. If Uzbek companies offer clients AI automation of call centers, "risk" turns into "product."
New occupations: what to learn
The WEF's forecast 170 million new jobs will concentrate mainly in: AI system deployment and management, data engineering and curation, AI security and auditing, cybersecurity, human-AI collaboration design. Narrow roles like "prompt engineer" are quickly becoming broader professions like "AI workflow architect."
An important change — the "shelf life" of skills is shrinking. A degree earned once is no longer a 40-year guarantee; continuous learning has become a condition of professional survival. Here Uzbekistan's world leadership in Coursera activity is a real advantage: a learning culture is already forming. The question is whether this activity is directed at the concrete skills the labor market demands — data literacy, productive use of AI tools, critical thinking.
Practical retraining paths are already visible: a call-center operator can become an AI conversation-scenario designer, a junior developer a code reviewer and system architect, a data-entry worker a data-quality controller. These transitions won't happen on their own — they need targeted short courses, employer-paid certification programs, and a qualification system that recognizes new roles.
Conclusion: practical steps for three sides
For young specialists: don't rely on a single narrow skill. If you're learning programming, learn not to write code with AI but to design systems, review code, and frame problems correctly — it's in that layer that human value persists. English and domain knowledge — finance, medicine, logistics — are not protection from AI but weapons for working with it.
For employers: the biggest mistake is seeing AI only as "cost cutting." Research shows successful companies don't cut jobs mechanically but redesign the work itself: routine tasks to AI, judgment, creativity, and client trust to humans. Investment in retraining employees is cheaper than finding new ones.
For the education system and policy: adapt curricula to the AI era — not memorization but problem-solving; not theory but practical projects. To soften the entry-level squeeze: internships and mentoring programs, support for AI adoption in small business, and an open data system tracking the labor market.
The main conclusion is simple: in the future jobs won't decrease — they'll change. The winners won't be those who fear AI but those who master it. Uzbekistan's young, learning-hungry population has a natural advantage in this race — it remains to be directed correctly.
