0
0
Somebody probably sent you the list. Or you read a headline about it, went looking for your own job, found it there, and felt your stomach drop.
Here is what actually happened. The US Bureau of Labor Statistics — the agency whose entire job is counting jobs — published a new way of sorting occupations by AI exposure. It placed 206 detailed occupations in the very highest tier. Customer service representatives are on it. So are web developers, public relations specialists, and a range of hotel roles.
The list is real. The reading of it that is spreading is not. And the agency that published the list said so — on the same page, in plain language.
What “AI Exposure” Actually Measures
The tiers do not rank jobs by how likely they are to vanish. They rank them by how many of the day-to-day tasks inside a job a machine can currently assist with or complete, compared with other jobs. That is a statement about tasks. It is not a statement about people, and it is not a statement about next year’s headcount.
The Bureau spells out the limits itself. Its own guidance says that “High” or “Very high” relative AI exposure does not necessarily mean employment will decline. Then it goes further and lists what the categories are explicitly not: not a forecast of employment growth or decline, not a wage forecast, not an estimate of how likely AI is to be adopted, not an estimate of worker replacement, not a productivity forecast.
Five separate disclaimers, attached to the thing at birth. That is not an agency hedging. That is an agency watching people misread a number and trying to get out in front of it.
The Numbers That Did Not Make the Headline
Here is where it gets genuinely odd. Take web developers, who sit in that very-highest-exposure tier. The Bureau projects employment for web developers and digital designers to grow 5 percent between 2025 and 2035 — faster than the 3 percent average across all occupations. Highest exposure tier. Above-average growth. Both true at once, from the same agency, in the same release.
That is not the whole picture, and it would be dishonest to pretend it is. Some jobs on that list are genuinely shrinking. Customer service representatives are projected to decline 5 percent over the same decade — a loss of roughly 141,800 positions. Office and administrative support as a whole is projected to fall 4 percent and shed 752,100 jobs, the largest decline of any major occupational group. Those are real numbers attached to real households.
But sit with one more figure, because almost nobody quotes it. Even while customer service representatives lose those 141,800 positions, the Bureau still projects around 289,500 openings in that occupation every single year of the decade. Every one of those openings comes from replacement — people retiring, or moving into different work.
An occupation can be shrinking and still need a quarter of a million people a year. “Declining” and “closed” are not the same word. The exposure list cannot tell the difference between them. Neither can a headline. You can.
What the Data So Far Actually Shows
Projections are estimates about a future nobody has seen. It is fair to ask what has already happened.
Economists at Stanford’s Institute for Economic Policy Research asked exactly that this year, in a policy brief titled “What is really happening to jobs? Separating AI hype from reality”. Their finding: since 2022, the unemployment rate for the most AI-exposed workers rose by 0.77 percentage points, while the rate for the least-exposed workers rose slightly more, by 0.85 percentage points.
Read that twice. The workers the exposure list would have you worry about most have, so far, done marginally better than the ones it implies are safest.
Those researchers are careful, and we should be too. They also found real evidence of a harder market for younger and less-experienced workers in exposed roles, and they say plainly that early evidence is not a promise about the long run. Their conclusion is not “AI will never cost anyone a job.” It is that there is little evidence AI is causing significant job losses right now. Those are different claims, and the gap between them is where most of the fear lives. It is the same gap that shows up when thousands of executives report no AI productivity gains and then confidently predict them anyway.
What to Actually Do With the List
Not much, if you were hoping it would tell you your future. But it is useful for one thing, and that thing costs nothing:
- Read it as a task map, not a verdict. Write down what you actually did at work last week, line by line. The list is about tasks — so look at yours. Which ones could a machine plausibly draft a first version of? Those are the ones to stop treating as the centre of your value.
- Notice what is left over. Judgement calls. Reading a room. Working out what the problem even is before anyone has named it. Being the person someone trusts with bad news. The exposure list measures none of that, which is why it runs out of things to say about you fairly quickly.
- Do not make a decision on a tier. The tier is relative. It only says your job contains more automatable tasks than some other job does. It carries no date and no probability. Fear is a poor planner, and it is worth knowing that a great many people are now staying in jobs out of fear rather than loyalty — which the list will happily feed if you let it.
None of that needs money, a course, or a career coach. It needs a sheet of paper and twenty honest minutes.
The Question the List Cannot Answer
Something worth noticing about this whole story: every measurement in it measures output. Tasks completed. Positions filled. Hours replaced. That is a fair thing to measure. It is also a remarkably small thing to be measured by.
There is an old idea, far older than any labour statistic, that a person is called good before doing anything at all — that worth gets assigned at the start rather than earned across a career and revoked in a restructure. It turns up in one of the oldest stories people have kept: God looks at a person who has not yet worked a single day, and calls that person good. Whatever you make of where that idea came from, it is doing something the exposure list structurally cannot do. It describes the person instead of the output.
An exposure score can tell you a great deal about your tasks. It has never once had an opinion about you.
If that is the question actually sitting underneath the panic — less “is my job safe” and more “what am I for” — we built a free What Is My Purpose? assessment for exactly that, and it takes about five minutes.
So Where Does That Leave You
The list will keep circulating. Somebody will forward it to you again, or you will find yourself checking it one more time at eleven at night.
When that happens, remember what you are actually looking at: a task-similarity ranking, published with five disclaimers by the very people who built it, containing a growing occupation in its highest-risk tier and a shrinking occupation with a quarter of a million openings a year.
It is a genuinely useful document. It is a terrible mirror.
Questions People Are Asking
What does “AI exposure” mean in the Bureau of Labor Statistics classification?
AI exposure in the BLS classification measures how many of an occupation’s tasks artificial intelligence can currently assist with or complete, relative to other occupations. It combines theoretical exposure (whether AI could do the task) with observed evidence of real AI use. BLS states the categories do not distinguish automation from augmentation and should not be read as forecasts of employment, wages, productivity, AI adoption, or worker replacement.
How many occupations are in the highest AI exposure tier?
The Bureau of Labor Statistics placed 206 detailed occupations in its highest (“very high”) AI exposure tier. Named examples include customer service representatives, web developers, public relations specialists, and a number of hotel roles.
Does high AI exposure mean my job will disappear?
No. The Bureau of Labor Statistics states directly that “High” or “Very high” relative AI exposure does not necessarily mean employment will decline. The ranking describes how many tasks within a job AI can touch, not how many people that job will employ. Web developers sit in the highest exposure tier and are projected to grow 5 percent between 2025 and 2035, faster than the 3 percent average across all occupations.
Which jobs are actually projected to decline the most?
In the Bureau of Labor Statistics 2025-35 projections, office and administrative support occupations show the largest decline of any major occupational group — down 4 percent, a loss of 752,100 jobs. Customer service representatives are projected to decline 5 percent, about 141,800 positions. Even so, that occupation is projected to have roughly 289,500 openings each year, all of them from replacing workers who retire or move into other work.
Is AI already causing job losses?
The evidence so far is limited. A 2026 Stanford Institute for Economic Policy Research policy brief found that since 2022, unemployment among the most AI-exposed workers rose 0.77 percentage points, while unemployment among the least-exposed workers rose slightly more, by 0.85 percentage points. The researchers concluded there is little evidence AI is causing significant job losses right now, while noting a harder market for younger and less-experienced workers in exposed roles, and cautioning that early evidence does not settle the long run.
What Do You Think?
If an agency ranked your job’s tasks tomorrow and handed you the number, would you want to see it — or would you genuinely rather not know? There is no right answer, and people split on this more evenly than you would expect. Tell us which way you land, and why, in the comments.
Share This
- 206 jobs just landed in the government’s top AI-exposure tier. One of them — web developers — is projected to grow faster than average. The list ranks tasks, not futures. https://bgodinspired.com/index.php/personal-growth-and-life-skills/ai-exposure-list-206-jobs/
- Customer service jobs are projected to shrink by 141,800 this decade — and still have around 289,500 openings a year. “Declining” and “closed” are not the same word. https://bgodinspired.com/index.php/personal-growth-and-life-skills/ai-exposure-list-206-jobs/
- I went looking for my job on the AI exposure list and found it. Then I read the five disclaimers the agency attached to its own list, and the panic didn’t survive them. Worth a read before you forward it to anyone. https://bgodinspired.com/index.php/personal-growth-and-life-skills/ai-exposure-list-206-jobs/
