Will AI Replace Jobs? What Studies and Official Data Say

Short answer: the major official studies do not predict that AI will wipe out jobs overall. Their shared finding is that the tasks inside jobs change more than whole occupations disappear. Some work, especially clerical and administrative roles, is clearly more exposed than the rest, and the effect on young workers is still being debated.

The World Economic Forum's Future of Jobs Report 2025, published in January 2025, projects 170 million jobs created and 92 million displaced between 2025 and 2030, a net gain of 78 million. Importantly, the WEF attributes this churn to technology, economic conditions, demographics and the green transition together, not to AI on its own.

Will AI replace jobs or just change them?

The International Labour Organization's global index, released with Poland's NASK in May 2025, estimates that about one in four workers is in an occupation with some exposure to generative AI. Only around 3.3% fall into the highest-exposure group. The ILO concludes that transformation is a more likely outcome than outright replacement.

The International Monetary Fund reached a similar headline in January 2024: roughly 40% of jobs worldwide are exposed to AI, rising to about 60% in advanced economies and around 26% in low-income countries. According to the IMF, about half of the exposed jobs could benefit from AI while the other half could be hurt by it.

What does AI exposure actually mean?

Most research measures exposure, not job loss. Exposure describes how much of a job's work AI could touch. That can mean automation, where AI does the task, or augmentation, where AI helps a person do it faster or better.

A highly exposed job is not necessarily a doomed job. A translator whose tasks are heavily exposed might end up editing and checking machine output rather than losing work entirely. Because each study defines exposure differently, their percentages cannot be compared directly, so it is worth checking who produced a figure and how.

Which jobs are most at risk from AI?

Across studies, a fairly consistent list emerges. The ILO identifies clerical and administrative work as the most exposed category. Other frequently cited areas include data entry, customer service, translation, writing and copywriting, bookkeeping, entry-level programming and parts of paralegal work.

Among the roles the WEF expects to decline fastest are cashiers, bank tellers, data-entry clerks and administrative assistants. The ILO also finds that women are more exposed than men, particularly in high-income countries, largely because they are more concentrated in clerical jobs.

Which jobs are safest from AI?

Jobs built on physical work, manual dexterity and in-person interaction show the lowest exposure. That includes construction, skilled trades, care work, nursing and other hands-on roles.

In fact, the WEF expects some of the largest growth in absolute numbers to come from farmworkers, delivery drivers, construction workers and care workers. Many of the jobs projected to add the most people this decade are not technology jobs at all.

What new jobs will AI create?

The WEF lists big data specialists, fintech engineers, AI and machine learning specialists and software developers among its fastest-growing roles. Cybersecurity, green energy and healthcare are also growth areas, and coverage of the report points to strong employer demand for technology and generative AI skills.

Newer titles such as AI trainer, AI engineer and AI governance or ethics specialist are also appearing. There is no official count of how many people hold these roles yet, so evidence for them is mostly anecdotal for now.

Keep in mind that the WEF figures come from a survey of more than 1,000 employers. They are projections of what companies expect, not measured outcomes, and the next edition of the report will revise them.

Is AI hurting young and entry-level workers?

This is the most contested question and the one most likely to shift as new data arrives. An August 2025 working paper from the Stanford Digital Economy Lab, using US payroll data, reported a relative decline in employment since late 2022 for workers aged 22 to 25 in highly exposed jobs such as software development and customer service. Older workers in the same roles held steady or grew.

An October 2025 analysis from the Yale Budget Lab, by contrast, found no clear economy-wide disruption to the US occupational mix since ChatGPT launched. Other factors, such as interest rates, may also explain part of the Stanford findings. For now, the early evidence is mixed rather than conclusive.

What skills should you learn to stay employable?

The WEF expects around 39% of workers' core skills to change by 2030. The skills employers rank highest include analytical thinking, resilience and flexibility, leadership, AI and big data, technology literacy, creative thinking, and curiosity and lifelong learning.

In practical terms, that means learning the AI tools relevant to your field, judging their output critically, and strengthening the people skills that machines handle poorly. Breaking your own job into tasks and asking which could be automated and which could be assisted is often more useful than worrying about the job title as a whole.