Data study · 1 October 2026 · 5 min read
AI vs robots: which one threatens your job?
Across 733 occupations, AI and robot exposure run in opposite directions and no job scores high on both. Which side your job is on, and which is closer.
AI and robots threaten almost completely different jobs. We compared Anthropic's new robot exposure data with its measure of how much each job's tasks are already done by AI models: across 733 occupations, the two run in opposite directions, and not a single occupation scores high on both.
If your job is mostly on a screen, the pressure comes from language AI. If it is mostly physical, it comes from robots, and much more slowly.
Two measures, one company
Both come from Anthropic's economists.
- Language AI exposure: the share of an occupation's tasks already showing up in automated, work related use of Claude, from the March 2026 labor market study.
- Robot exposure: a 0 to 3 index of how much of an occupation's physical work robots have been shown to do, and in how messy a setting, from the September 2026 robotics study.
We matched the two on occupation code. The rank correlation is negative, minus 0.31: the more exposed a job is to one, the less it tends to be exposed to the other. Not one of the 733 occupations scores at least 1 on robots and 25 percent on language AI.
Where each one lands
| Occupation | Language AI exposure | Robot exposure (0 to 3) |
|---|---|---|
| Computer Programmers | 74.5% | 0.00 |
| Customer Service Representatives | 70.1% | 0.00 |
| Data Entry Keyers | 67.1% | 0.10 |
| Accountants and Auditors | 34.8% | 0.02 |
| Lawyers | 16.7% | 0.00 |
| Registered Nurses | 6.0% | 0.01 |
| Cashiers | 8.5% | 1.00 |
| Construction Laborers | 2.8% | 1.05 |
| Laborers and Freight Movers | 0.0% | 1.46 |
| Heavy and Tractor-Trailer Truck Drivers | 0.0% | 1.51 |
| Taxi Drivers | 0.0% | 2.22 |
Data from Anthropic's Economic Index releases: labor_market_impacts/job_exposure.csv and robot_exposure/robot_exposure_occupations.csv.
The top of the table is office work with no physical component at all. The bottom is driving and handling work that no language model can do. Nurses score low on both.
Which one is closer
Language AI, by a wide margin. Its exposure is measured from what people already use it for at work, and a finished task costs a few dollars at most. Robot exposure measures what robots can do somewhere, and the same study found robots are cheaper than a person for only 0.3 percent of physical tasks today.
The 2026 job cuts that cited AI were concentrated in office functions: customer service, data operations, entry level software and finance back offices. Driving and warehouse work were not among them.
Adding them together
Anthropic's own robotics paper does the sum: about half of all US work is exposed to language AI alone, and about 81 percent once robots are counted. That sounds like almost everyone. It is really two separate groups of workers, facing two very different timelines.
Our index scores AI displacement risk from published research, most of it about language AI, so office roles sit higher than drivers: Customer Service Representatives (68) against Taxi Drivers (25). For the short list of jobs that sit low on both measures, see jobs safe from AI and robots.
Which side is your job on?
Most jobs mix screen work and physical work, and the mix is what decides your exposure. The assessment asks how your week is actually spent and scores it in about two minutes, free and without signing up.
Where does your job sit?
Every number above is an occupation average. Your own exposure depends on your seniority, your sector and how much of your day is routine. Answer a few questions and get your personal score free.
Get my personal scoreFree · No credit card · No signup
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.