Guide · 31 July 2026 · 6 min read

Is your job at risk from AI? How to tell

Job titles and industries are weak signals. Across 961 scored occupations, what predicts exposure is your task mix, not your title.

The short answer: your exposure depends on how much of your week is spent producing and processing text and structured documents, not on your job title or your industry. The larger that share, the more exposed you are, and two people with identical titles can sit far apart on it.

That is worth spelling out, because most people answer this question a different way. They pattern match on the title. They read that AI writes code, or drafts contracts, or answers support tickets, and they decide whether their own title sounds like one of those things.

That instinct is understandable and it is close to useless. Across 961 occupations scored against 18 published research sources, the signal that separates an exposed role from a protected one is not the title and not the industry. It is the mix of tasks inside the day.

Here is how to read your own job the way the research reads it.

Start with your tasks, not your title

Every source in our base that works at the task level lands in the same place. Exposure is a property of activities, not of job names. A title is a label a company chose. The tasks underneath it are what a model either can or cannot do.

So the useful exercise is to break your week into the things you actually do, then sort them into three groups.

Things that are mostly language in and language out. Drafting, summarising, rewriting, translating, formatting, first pass research, routine correspondence, filling structured documents from unstructured notes. This is where current systems are strongest, and it is the group that drives almost every high score in our index.

Things that need judgment under accountability. Deciding what to recommend when the evidence conflicts, choosing what to escalate, taking responsibility for a call that could go wrong. Models can draft the analysis behind these. Someone still has to own the decision, and in many fields that ownership is a legal or professional requirement rather than a preference.

Things that need a body in a place. Physical work, unpredictable environments, hands on care, anything where the work happens to objects or people rather than to documents.

Count roughly what share of your week sits in the first group. That fraction, more than anything on your business card, is what the research is measuring.

The same industry contains both extremes

If sector were the driver, occupations inside one industry would cluster. They do not.

Medical Records Specialists score 60, among the most exposed occupations in the whole set. Meanwhile Orderlies score 8. Same industry, 52 points apart.

That is the whole argument in one comparison. Two roles, same industry, same regulatory environment, often the same building. The gap between them is task content and nothing else.

Sector means still tell you something, but they are averages over very wide ranges, and the range matters more than the middle:

Read that table as a starting prior, not a verdict. Knowing your sector narrows the guess a little. Knowing your tasks narrows it a lot.

Seniority changes the answer more than people expect

Two people can hold the same title and sit far apart, because the share of the week in that first group is not fixed across a career.

Junior roles in knowledge work tend to concentrate exactly where models are strongest. The first draft, the research summary, the formatting pass, the routine ticket. That is not an accident of how companies are organised, it is close to the definition of an apprenticeship in office work: you do the parts that are teachable and checkable while you learn the parts that are not.

That is why an occupation average can be misleading in both directions. A senior person in an exposed occupation may spend most of the week on judgment and accountability. A junior person in a moderately scored occupation may spend most of it on exactly the tasks that automate first. We wrote separately about what the payroll data shows for early career workers, which is the sharpest evidence on this point so far.

What a high score does not mean

Our scores measure task overlap between an occupation and current model capability. That is a narrower claim than it sounds, and worth being blunt about, because the gap between the two is where most misleading coverage lives.

A high score does not mean the occupation disappears. It means a meaningful share of its current tasks are within reach of the tools. What happens next depends on adoption, cost, regulation, liability and whether the freed up time gets spent on more of the same work or something else. None of that is in the score.

It is also worth keeping the absolute numbers in view. The mean score is 30. The median is 29. Only one of 961 occupations clear our high risk threshold of 70. The honest reading of the aggregate research is not that whole professions are about to vanish. It is that the composition of a lot of jobs is shifting, and that the shift is fastest in text heavy work.

A short checklist

If you want a rough answer in five minutes, ask these in order.

1. What share of my week is producing or processing text and structured documents? Higher share means higher exposure.
2. Of that share, how much is a first draft or a routine pass that someone else checks anyway? That is the most exposed slice of all.
3. Does my role carry accountability that a person has to sign for? That is the most durable part of most knowledge jobs.
4. Does the work require being physically present, or handling unpredictable physical situations? That is the strongest protection in the current data.
5. Am I early in the career ladder in a text heavy field? If so, treat the occupation average as a floor on your exposure, not a description of it.

Then check the occupation itself. Our index has a page per role with the score, the sources behind it, and the spread between them, and the spread is often the most informative part. You can browse the full ranking, look up your occupation, or read how the scoring works before you decide how much weight to give any of it.

Figures come from the AI Job Risk Index and were current when this was published. Scores change as new research is ingested, so the index is always the live version. See how scoring works. Informational guidance based on published research, not professional career or financial advice.