Skip to content
MaungaAI news, explained

Is it true?

Is it true that AI will wipe out half of all jobs?

You'll learn where the idea that AI will eliminate half of all jobs comes from, what the studies people cite actually measure, and what today's employment data shows.

Published · Spanish original

Translated with AI from the Spanish original. Nobody has reviewed this translation; if anything seems off, the Spanish version is the reference.

No. No serious study concludes this. The famous 47% figure measured which jobs could be automated in theory, not expected layoffs, and current data shows no mass job losses. The future, however, is still open.

Where the claim comes from

The phrase has a fairly specific origin. In 2013, two researchers at the University of Oxford, Carl Benedikt Frey and Michael Osborne, published a study estimating that 47% of jobs in the United States were at high risk of automation, perhaps within one or two decades. That "almost half" has been repeated ever since in headlines, talks and on social media.

To reach that figure, the authors looked at 702 occupations and estimated how far technology could do the typical tasks of each one. If an occupation came out as highly automatable, all of its jobs counted as "high risk".

The idea came back strongly after ChatGPT arrived. In May 2025, Dario Amodei, head of Anthropic, told Axios in an interview that AI could eliminate half of all entry-level office jobs within one to five years. He also said that, over the same period, unemployment in the United States could rise to between 10% and 20%.

This is a prediction by someone who runs an AI company, not a fact. It also refers to a specific group of jobs, not all of them.

The strongest version of the argument

People who hold this view have reasons worth taking seriously. Today's AI programs already write texts, summarise documents, translate, write code and answer queries. These tasks fill a large part of the working day for millions of office workers.

On top of that, this kind of software can reach many companies at once and at low cost, while a physical machine has to be built and installed. If one person with AI does the work of two, a company could hire fewer people. The risk is real and shouldn't be brushed off.

What the data says

The first important point: the Oxford study did not predict that half of all jobs would be lost. It measured how much work could be automated in theory, with the technology that was on the horizon. It didn't consider whether companies would find it worthwhile, how long it would take, or the new jobs that would appear.

When other researchers redid the calculation looking at tasks instead of whole occupations, the figure dropped sharply. A study published by the OECD in 2016, by Melanie Arntz, Terry Gregory and Ulrich Zierahn, estimated that on average only 9% of jobs in 21 rich countries were automatable. The reason is that almost every job mixes tasks that are easy to automate with others that aren't.

Studies on the newest AI, the kind that writes and holds conversations, mostly use the word "exposure". An exposed job is one where AI can do, or help do, some of the tasks. That doesn't mean it will disappear: it may change, get faster or require different skills.

Using this idea, the International Monetary Fund estimated in January 2024 that around 40% of jobs worldwide are exposed to AI. In rich countries the figure rises to 60%, and of those exposed jobs, roughly half could benefit from AI because it would help people work better; in the other half, demand for workers, wages or hiring could fall.

In 2023, the International Labour Organization published a global analysis of AI that generates text and images. It concluded that its most likely effect is to complement work, that is, to change some of the tasks, rather than replace entire jobs. Clerical jobs came out as the most exposed.

In May 2025, the ILO refined its estimate: one in four workers worldwide is in an occupation with some exposure, but only 3.3% are in the highest category. It still holds that jobs are most likely to change, not disappear.

The World Economic Forum surveyed more than a thousand large companies for its 2025 Future of Jobs report. Based on their answers, it estimated that between 2025 and 2030 about 170 million jobs will be created and about 92 million will disappear, a net gain of about 78 million. There are two caveats. These are the companies' own forecasts, not data. And they don't only cover AI, but also other changes, such as the economy or demographics.

What we can already see in the job market

Beyond forecasts, some studies have looked at what has already happened. In October 2025, Yale University's Budget Lab compared how the mix of occupations in the United States had changed since ChatGPT arrived. It found no broad disruption of the job market. The pace of change was similar to that of other technologies, such as the arrival of the internet. Its monthly updates, the latest of which we checked uses data from August 2026, still don't show one.

There is a more worrying sign. In August 2025, researchers at Stanford University analysed US payroll data, and in August 2026 they updated the analysis with data up to June 2026. In the occupations most exposed to AI, employment of young people aged 22 to 25 is 19% lower than that of young people in less exposed occupations, compared with November 2022.

The gap has been growing, and it is mainly because fewer young people are being hired, not because of layoffs. There is no such gap among more experienced workers. The authors warn that they are describing a pattern and do not show that AI is causing it.

Taken together, this data paints a different picture from mass layoffs. AI may be making it harder to land a first job in some office roles, without, for now, any widespread loss of jobs.

What we don't know

Almost all of these studies are from the United States or make global estimates. There is much less information on how AI is affecting jobs in Spain or other European countries.

It's also unclear how much of what we see is due to AI. Young people are also affected by interest rates, cutbacks at tech companies and the overall economy, and separating each cause is difficult.

And the big question remains open: what will happen in ten years. AI is improving fast, and the fact that there have been no mass layoffs so far doesn't guarantee there won't be. No one can say for sure that half of all jobs will disappear, or that none will be lost.

Watch out: When you read that "X% of jobs are at risk" or "exposed to AI", keep in mind that it almost always means AI can do some of the tasks, not that those jobs will disappear. These are very different things, and mixing them up is behind many alarmist headlines.

Sources

More guides

All guides