It is natural to talk about exposure by sector. Creative work is exposed. Trades are safe. Transport is somewhere in between. The group medians in this index support that kind of talk, up to a point.
The point where they stop supporting it is here:
| Major group | Most exposed occupation | Least exposed occupation | Spread |
|---|---|---|---|
| Arts, design and media | Technical Writers, 68.9% | Athletes and Sports Competitors, 8.4% | 60.4 points |
| Transportation | Traffic Technicians, 62.4% | Industrial Truck and Tractor Operators, 4.1% | 58.3 points |
| Construction | Energy Auditors, 56.2% | Stonemasons, 0.0% | 56.2 points |
| Personal care | Concierges, 60.1% | Embalmers, 5.3% | 54.7 points |
| Sales | Telemarketers, 73.3% | Models, 19.9% | 53.4 points |
The distance between the most and least exposed occupation inside a single family is routinely larger than the distance between the most and least exposed family.
Construction is the sharpest case
Construction has the lowest median of any major group at 5.1%. It is the group people name when they want an example of work that is out of reach.
It also contains Energy Auditors at 56.2%, which is higher than the median for management, education, or life and physical science. An energy auditor's work is inspection, modelling, calculation and report writing. It sits in the construction family because of where it happens, not because of what it is.
Meanwhile Stonemasons score exactly zero. Not low. Zero. Every task in that occupation maps to activities that current systems cannot produce.
Both of those are construction.
Why sector-level talk misleads
Occupational families are administrative groupings. They cluster jobs by industry, setting and tradition, not by what the work consists of at the level a machine either can or cannot do it.
Exposure tracks something else entirely: whether the output is symbolic, whether a body has to be there, whether a licence attaches, how much private context is needed, and how expensive an undetected error is. Those properties cut across sectors at right angles.
This is the practical reason the index publishes every occupation separately and publishes each occupation's tasks underneath it. A sector average is a fact about a filing system. It is not a fact about anybody's job.
How to use this
If you want to know where your own work sits, the sector is the wrong unit and so, often, is the job title. The right unit is the task list. Two people with the same title in the same industry can have materially different exposure depending on which parts of the role they actually do, and the task table on every occupation page is there to be read that way: not as a verdict on the occupation but as a list you can go through and tick.
Figures from release v2026.Q3. Spreads are computed across occupations within each SOC major group with more than five scored occupations. See the methodology page.