Call · 30 min
AI systemsMattia Esposito8 November 20267-minute read

Will AI take my job? The figures point to jobs that change, more than jobs that disappear.

Whether artificial intelligence will take people's jobs is a question with a measured answer, and it comes from two international institutions. Here are their figures, the most exposed occupations and the protections Italy already has.

In brief

Exposed doesn't mean replaced. A job is exposed when part of its tasks can be done by generative AI; most jobs are still made up of tasks that need a person.

According to the International Labour Organization (ILO), in its report of 20 May 2025, one worker in four worldwide is in an occupation with some exposure to generative AI, but only 3.3% are in the highest exposure category.

In Italy two laws already protect workers: Law 132/2025 requires employers to tell workers when AI is used, and Legislative Decree 179/2026 bans decisions about employment taken by an automated system alone.

Written by: Mattia Esposito, AI and marketing consultant, founder of Itria AI. From Puglia, based in Turin since 2023, he works remotely in Italy and abroad. LinkedIn profile.

This page is about the work of people who use AI or feel its effects, not about how to build a system. For that there are the human in the loop entry and the page on AI agent or chatbot.

What the figures say: one worker in four

The ILO estimates that one worker in four worldwide is in an occupation with some exposure to generative AI, and that 3.3% of employment is in the most exposed category. The index, published in Working Paper 140 on 20 May 2025, scores 29,753 occupational tasks one by one, using worker surveys, experts and AI models.

The authors' conclusion fits in one line: “transformation of jobs is the most likely impact of GenAI”. The reason is that most jobs are made of tasks that still need a person.

Which jobs are most exposed

Clerical jobs remain the most exposed, according to the ILO, and highly digitised roles in technical and professional occupations are catching up. Exposure also varies with a country's income and with gender, because women and men on average do different jobs.

DataValueSource
Workers with some exposure

One in four, worldwide

ILO, May 2025

Highest exposure category

3.3% of global employment

ILO, May 2025

Women in the highest category

4.7%

ILO, May 2025

Men in the highest category

2.4%

ILO, May 2025

Exposure in high-income countries

34% of employment, against 11% in low-income countries

ILO, May 2025

Most exposed occupations

Clerical work

ILO, May 2025

Employment exposed to AI, worldwide

Almost 40%

IMF, January 2024

Employment exposed in advanced economies

About 60%

IMF, January 2024

The IMF and ILO figures don't add up: they measure different things, AI in general for the first and generative AI for the second, with different methods. Read together, they say exposure is wide and that its effect depends on how AI enters the tasks.

Exposed doesn't mean replaced

The International Monetary Fund splits the exposed jobs in advanced economies into two halves. In its article of 14 January 2024 it says roughly half may benefit from AI integration, with higher productivity; in the other half AI may carry out key tasks now done by people, with effects on pay and hiring.

The difference between the two halves lies in which tasks pass to the machine. Where AI prepares and a person decides, the job shifts towards checking and relationships; where the whole task is automated, the job really changes. That's why the useful question is about a job's tasks, one by one, more than the job as a whole.

What changes in the tasks: three examples

The same job splits into tasks AI can prepare and tasks that stay with a person. Three examples from an ordinary business show where the line runs. They're not estimates: they're a way of reading the figures above against your own work.

In accounts, AI can read incoming invoices, fill in the fields in the business software and draft payment reminders. What stays with a person is checking the exceptions, phoning the customer who doesn't pay and deciding what to do about them.

In a sales office, AI can draft the reply to an enquiry, look up the price in the price list and summarise a customer's history. What stays with a person is the price to offer, the negotiation and the promise that commits the business.

In a professional practice, AI can summarise long documents and prepare a first draft. What stays with a person is the judgement on the case, the responsibility towards the client and professional confidentiality. In all three cases the work moves from doing to checking, and the person checking has to know what they're looking at.

In Italy: businesses and the protections already in place

In Italy, AI in business is still a minority, but it doubled in a year. According to Istat, 16.4% of companies with at least 10 employees used at least one AI technology in 2025, up from 8.2% in 2024.

Article 11 of Law 132/2025 says the use of AI at work must be safe, reliable and transparent, and requires employers to inform workers in the cases set out in Legislative Decree 152 of 1997. The page on Law 132/2025 covers the rest.

Article 41 of Legislative Decree 179/2026, in force since 23 October 2026, adds that decisions on hiring, changing or ending employment, including disciplinary measures and performance reviews, are not taken by automated processing alone. The final decision rests with a person, and the worker's consent is not enough to lift the ban.

If your job is among the most exposed

The first step is to go through your tasks one by one and see which ones an assistant can prepare. Drafting, summarising, filling in, checking data: these are the tasks that change first. The work that remains, and carries more weight, is deciding, checking and dealing with people.

The second step is learning to use the tools the company chooses, under the rules the company writes. The AI Act requires anyone using AI systems for work to look after their staff's skills, as explained on the page about mandatory training, and tools used behind the company's back are the case of shadow AI.

The third step is about the value of your work. The tasks that remain, checking a result, understanding a customer, deciding an exception, are worth more precisely because the repetitive ones pass to a system. People who do them well become the person who approves, and Italian law makes that role mandatory for decisions affecting workers.

For people running a business

The choice that matters is where to put the person. A system that prepares and a person who approves changes the work without taking it away: the person who did the task by hand now checks it and answers for it. It's also the simplest way to comply with the two Italian laws on work.

Before bringing in a system, it pays to tell the people who will use it, explain what changes in their tasks and write down who approves what. The rule is described in the human in the loop entry.

It's also worth measuring before and after: how long a task took, how long it takes with the system, how many errors reach the person approving. Those are the figures that let you decide where AI really helps and where it doesn't, and tell people with facts rather than promises.

Questions and answers

Will artificial intelligence take people's jobs?

According to the ILO, the most likely effect is that jobs are transformed: one worker in four worldwide is in an occupation with some exposure to generative AI, but only 3.3% are in the highest exposure category.

The International Monetary Fund estimates that in advanced economies about half of exposed jobs may benefit.

Which jobs are most at risk from AI?

For the ILO the most exposed are clerical occupations, and exposure is growing for some highly digitised roles in technical and professional occupations.

Exposed means that part of the tasks can be done by AI, not that the job disappears.

How many jobs are exposed to artificial intelligence?

The International Monetary Fund estimates almost 40% of global employment, and about 60% in advanced economies.

The ILO, which measures generative AI only, estimates one worker in four with some exposure.

Does an employer have to say they use AI?

Yes. Article 11 of Law 132/2025 requires employers to inform workers about the use of AI in the cases set out in Legislative Decree 152 of 1997.

And Legislative Decree 179/2026 prohibits decisions about the employment relationship being taken by an automated system alone.

Can an algorithm fire a worker in Italy?

No. Article 41 of Legislative Decree 179/2026, in force since 23 October 2026, requires that decisions on hiring, changing or ending employment are not taken by automated processing alone.

The final decision rests with a person, and the worker's consent is not enough to lift the ban.

Can AI replace an office employee?

Rarely in full: according to the ILO, Working Paper 140 of 20 May 2025, clerical jobs are the most exposed to generative AI, but the most likely effect is that jobs change, because most tasks still need a person.

The International Monetary Fund splits exposed jobs in advanced economies into two halves: in one, AI raises productivity; in the other, it can take over key tasks. The useful question is about tasks one by one: where AI prepares and a person decides, the job changes without disappearing.

Notes on sources

  1. International Labour Organization, Generative AI and jobs: A refined global index of occupational exposure, Working Paper 140, 20 May 2025, read on 8 November 2026: one in four, 3.3%, 4.7% and 2.4%, 34% and 11%, clerical occupations; the 29,753 tasks from Table 1 of the PDF. The quotation comes from the summary.
  2. International Monetary Fund, AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity, 14 January 2024.
  3. Istat, Imprese e ICT (ICT in enterprises), 2025, covering companies with at least 10 employees.
  4. Law No 132 of 23 September 2025, Article 11, and Legislative Decree No 179 of 7 October 2026, Article 41, current texts on Normattiva read on 8 November 2026 (in Italian).
  5. This page is a survey, not legal advice or a forecast about your job.
·The next step

A system that prepares and a person who decides. The work changes without disappearing.

The first step with Itria is a thirty-minute video call: we look at which tasks an AI system can prepare and where a person needs to stay in place before anything is sent. Drop us a line about what's slowing you down.