AI will replace most white-collar jobs within a decade

What's this about?

People disagree about whether AI will replace most office jobs within the next ten years.

The key question is whether AI will remove whole jobs or just change parts of them.

What supporters say

• Many office jobs include tasks that AI may do, such as writing, filing, and checking data.
• Young workers already show warning signs, with fewer jobs in fields where AI can do many tasks.
• Better, cheaper, and safer AI could soon handle more skilled work than it handles today.

What critics say

• AI may change some tasks without taking away the whole job.
• Early studies often show AI helping workers, rather than fully taking their place.
• In the past, new machines removed some tasks but also created new work and jobs.

How to read this

The number of points on each side does not show who is right; strong proof matters more than a long list.

The bottom line

AI will likely change many office jobs by 2036, especially jobs for young workers.

But current facts do not show that AI will remove most office jobs within ten years.

AI is likely to change many white-collar jobs over the next decade, but current evidence does not show that it will eliminate most of them by 2036. The strongest findings point to widespread task disruption and particular risks for younger workers, not proof of majority job loss.

The case for

White-collar work is unusually exposed to AI. The International Labour Organization identifies clerical occupations as especially vulnerable, while also finding growing exposure in some professional and technical fields. IMF estimates suggest that about 60% of workers in advanced economies are in occupations with high AI exposure (see Figure 4). Administrative, legal, engineering and other professional activities may all include tasks that future systems could perform. 1

But exposure measures show the share of work that AI might affect, not the number of jobs that will disappear. They depend on assumptions about what systems can do, how reliable they are and whether companies adopt them.

There are also early signs of disruption among younger workers. A Dallas Fed summary found that employment among 22- to 25-year-olds in the most AI-exposed occupations had fallen 13% since 2022. Stanford researchers reported that employment for this group was about 19% below a comparison trend, while finding no similar gap among more experienced workers. 2

That pattern could indicate fewer entry-level openings, as companies automate or restructure junior tasks. It is stronger evidence of concentrated disruption than of economy-wide replacement. Still, it suggests that AI may affect how people enter white-collar careers before it eliminates large numbers of established roles.

AI could also spread beyond routine clerical work. If systems become more capable, reliable and affordable, automation may reach a wider range of professional tasks than earlier technologies did. 3 The scale of that expansion will depend on regulation, computing costs, business decisions and whether rising productivity creates enough additional demand to preserve jobs.

The case against

Being exposed to AI does not mean an entire occupation will vanish. Exposure indexes usually measure individual tasks, while jobs combine many activities. Judgment, accountability, interpersonal skills, contextual knowledge and newly created responsibilities may remain difficult to automate. The ILO says generative AI is more likely to transform jobs through assistance than fully automate them, and the IMF distinguishes between occupations where AI may replace workers and those where it may make them more productive. 4

Workplace studies so far more often show augmentation than mass replacement. In a study of 5,179 customer-support agents, access to a generative-AI assistant increased the number of issues resolved per hour by 14% on average, and by about 34% for novice and lower-skilled workers. The study found higher productivity in that setting, not the removal of workers. 5

History also offers reasons for caution. Automation can displace people from existing tasks while creating new tasks, occupations and sources of demand. A World Economic Forum survey of employers predicted 170 million jobs would be created and 92 million displaced worldwide by 2030. That forecast covers all types of work, reflects employer expectations and is not specific to white-collar jobs, so it cannot settle this claim. 6

The early-career employment figures also have several possible explanations. Delayed hiring, changing job requirements or the replacement of junior tasks could all produce the observed pattern while experienced workers become more productive. None proves that more than half of white-collar jobs will disappear.

The bottom line

The evidence favours a more cautious conclusion than the claim makes. Confidence is low that AI will eliminate more than half of existing white-collar jobs by about 2036. Confidence is substantially higher that AI will affect many white-collar tasks and cause serious, uneven disruption, especially for junior workers.

The central unknown is how task-level AI capability will translate into company decisions to replace workers. That outcome will be shaped by adoption costs, reliability, regulation, demand growth, new tasks and the balance between using AI to assist employees and using it to reduce headcount. Current studies measure different things—task exposure, productivity, employment patterns and employer expectations—and cannot be combined into a reliable forecast of majority job elimination. The claim is therefore not established, even though significant white-collar upheaval is plausible.