Will AI take my job?

Most weeks there's a headline about artificial intelligence (AI) and jobs. Some say it will replace millions of workers. Others say it will create millions of new jobs. Both kinds of headline are reporting something real, which is part of why the question feels so hard to answer. So let's take it slowly. The answer is more useful than either headline, and a lot less frightening.

Start with what a job is. Almost no job is one task. A nurse checks charts, talks to worried families, notices when a patient seems off, and fills in a lot of paperwork. An office manager answers email and calms down whoever is upset. When people study AI and work, they mostly look at tasks, not whole jobs. AI is good at some tasks, like drafting, summarizing, and sorting. It's much weaker at others, like reading a room, earning someone's trust, or being the person who answers for a decision.

The largest regular survey of employers is the World Economic Forum's Future of Jobs Report. Its 2025 edition asked more than a thousand large employers what they expect between now and 2030. Worldwide, they expect:

• About 170 million new jobs created.
• About 92 million jobs displaced.
• A net gain of about 78 million.

Those are employers' forecasts, and forecasts can be wrong. They also count every cause of change, from the economy to climate, and AI is only one of them. Still, they point in a clear direction: a lot of movement, in both directions at once.¹²

The same report expects some roles to shrink, especially clerical work like cashiers and administrative assistants, where much of the day is routine information handling. It expects others to grow, from technology jobs to care work such as nursing, and frontline jobs such as farm work, delivery, and construction. And it expects the skills inside most jobs to shift. Employers expect about two in five of the skills workers rely on today to change by 2030.²³

But the net number isn't the whole story. A gain around the world doesn't help the person whose particular job shrinks, and changes like this are hard on the people they land on. That's why retraining and good policy matter, and why it's fair to expect employers and governments to take them seriously. If you're worried about this, that's a sign you're paying attention.¹

The most useful move is small, and it starts with your own week.

• First, write down the tasks that fill a normal week.
• Second, pick one that is slow and dull, try an AI tool on it, and check the result.
• Third, mark the tasks that need your judgment, your relationships, or your name on them, and keep investing there.

That's what augment, not replace, looks like in practice. The tool takes some of the routine, and the parts that need you get more of your time.

As AI writes more first drafts, the work that matters more is checking them. Is this right? Is it fair? Would I put my name on it? That's a human skill, and it gets stronger with practice. People who know what these tools can and can't do are in a much stronger position than people who don't, and you're becoming one of them.³