Will AI take my job? What the BLS data says at 16

The US Bureau of Labor Statistics now sorts jobs by AI exposure. It says plainly that this is not a forecast. Here's how to use it, and what to read beside it.

Will AI take my job? Nobody can tell you that for a specific job, and the US Bureau of Labor Statistics (BLS) doesn't pretend it can. Its new AI exposure categories sort jobs as Low, Moderate, High or Very high. They aren't employment forecasts, and they don't measure AI's impact on employment.^[1] At 16, think of them as a way to ask better questions, not as a verdict.

What did BLS actually publish?

On 27 August 2026, alongside its 2025–35 projections, BLS released supplemental data on AI exposure.^[2] It covers 831 detailed occupations and puts each one in one of four categories.^[1]

Five outside studies feed the sorting. Three estimate how well AI capabilities match the abilities and tasks of a job. Two look at what people have actually done with two AI chat tools.^[1]

The two ends of the scale are easy to picture. In a Low job, the requirements match poorly with current AI, and chat-style AI hasn't been seen doing many of its tasks. In a Very high job, AI can do or assist a larger share of tasks than in other jobs.^[1]

What do the categories not tell you?

This is the part most headlines skip. BLS lists the limits itself:

• The categories compare jobs with each other. They aren't an absolute level.^[1]
• Exposure "does not imply job loss, productivity gains, automation probability, or wage effects."^[1]
• They don't separate AI that replaces a task from AI that helps a person do it.^[1]
• The theoretical sources reflect AI capabilities available no later than mid-2023, and most of them look at language tools, not image or video.^[1]

So a High label can mean a job where AI is a daily tool. A Low label doesn't mean safe. Either way, you're looking at a snapshot.

Does AI exposure match job growth?

Not neatly, and that's the useful lesson. BLS credits AI with part of the growth in computer and mathematical jobs, and it projects data scientists to grow 34.6 percent.^[2] In the same release, office and administrative support is the fastest-declining group, and the release names automation tools, including AI, as one reason.^[2] BLS also calls the uncertainty about AI's impact very high, and it applied its AI adjustments conservatively.^[2]

Nurse practitioners and solar installers sit at the top of the growth list. So do data scientists. They have little in common except that employers want them.

What does this mean if you're 16?

Here are four things you can do with all this.

1. Think in tasks, not job titles. The categories are built from tasks and abilities.^[1] Ask yourself which tasks in a job you'd enjoy, and which ones a tool could take over. Most jobs are a mix.
2. Read exposure and growth together. A job can be highly exposed and still growing, or less exposed and shrinking. One number never settles it.
3. Don't let a label choose for you. You'll work for decades, and the figures stop at 2035. The next projections, for 2026–36, are due in 2027.^[1]
4. Start from what you like doing. Your interests tell you which families of work to look at first. Then use exposure to plan which skills to add, such as working with the tools, not against them.

If you're a parent, the same goes for your family. Use the data to talk about tasks and interests, not to rule a job out.

Where to look next

Open the BLS table yourself. It lists every occupation's category and its 2025–35 projection, and the BLS page links to it.^[1] Look up two or three jobs that already interest you instead of scanning all 831.

Not sure which jobs to look up? The CAT test is 23 questions, takes about 10 minutes and is free. It's built on the RIASEC model, so it gives you ideas to reflect on, not a definitive answer. After that, What should I study? can help you turn those ideas into a shortlist.

← All news