EXPOSED
The modal worker here is a local delivery driver — van or box truck, dozens to hundreds of stops a day — and the hard part is physical: getting a package from a curb, up steps, past a gate or a locked lobby, to a specific door. Language AI can't touch that, and driverless last-mile vehicles still can't do the walk-up portion at all. The exposure is that route planning, load sequencing, dispatch, proof-of-delivery, and customer notifications are already software, no CDL is required for most of these jobs, and pay pressure plus driver-assist tech means the job gets more monitored and more standardized rather than safer.
Headcount grew steadily across the period.
Median pay $34,730 → $44,860 +3.3% in real terms
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing. The marked year is 2020.
BLS projection, 2024–2034
+7.3% 1,079,800 → 1,158,600 on the projections basis
Growing, and only partly exposed
The BLS expects +7.3% more of these jobs by 2034, and at 46/100 the work is only partly exposed — some tasks are automatable, the core of the job is not. Nothing here is in tension.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~120,200 openings a year on average, including replacing people who leave.
DriverRoutemanServicerDelivererCar EscortVan DriverRoute RiderLocal DriverOrder WorkerParts DriverParts RunnerRoute DriverStore DriverTruck DriverWagon PersonErrand RunnerPick Up DriverDelivery DriverDelivery PersonRoute DelivererService ProviderWarehouse DriverCommercial DriverDirectory Carrier
Holding it up: embodiment . Weakest point: liability shield .
Tasks largely resist digitisation At 14 the driving itself is genuinely contestable — highway and depot-yard segments are the parts autonomy attacks first — but the day is dominated by the 100-200 stop walk-up cycle: parking illegally-but-tolerated, hauling a 40lb box up a stoop, finding the side entrance, dealing with a dog or a locked vestibule, and that sequence has no software substitute, which is why it sits above the mixed band rather than at 18 alongside work with no digitisable core.
Hands-on in uncontrolled environments 17 reflects that the workspace is the street in whatever weather it offers — ice on loading ramps, unlit apartment complexes, double-parking in traffic, lifting and carrying all day — and the only reason it isn't 19-20 is that the vehicle cab is a semi-controlled environment and the loads are consumer-scale, not rigging or trenching.
No licence, no signature requirement 3 because a class C licence and a clean-enough MVR is the whole entry requirement for vehicles under 26,001 lbs — no CDL, no DOT medical card for most local work, no professional body — so the employer's insurer and the company DOT number absorb the risk and nothing about you is legally hard to replace.
Executes defined procedures on defined inputs 6 fits work where the calls are real but small and bounded — reattempt or return to station, leave it at the door or take it back, this street is flooded so take the next one — all inside a scan-and-scope framework where the handheld dictates sequence and a dispatcher resolves anything genuinely ambiguous.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (14/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (3/20) is whether the law requires a licensed human to sign. Trust premium (6/20) is whether buyers specifically pay for a person. Judgment and accountability (6/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 15 of this occupation's 46 points (33%).
Embodiment (17/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
Locomotive Engineers EXPOSED
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 63/100, still EXPOSED.
Growth of the segments where the buyer is paying for an identified, background-checked, named person: pharmacy and specialty-drug courier, clinical specimen transport, white-glove furniture/appliance delivery with placement and haul-away, in-home installation. These are contracted at multiples of parcel rates specifically because a vetted human enters the premises. A shift of the occupation's mix toward these raises trust premium even with no regulatory change.
Expansion of statutory personal-verification duties onto the driver: state alcohol-delivery statutes requiring the deliverer to check ID and refuse intoxicated recipients, DSCSA chain-of-custody handoffs for pharmaceuticals, and specimen/controlled-substance courier rules. Where the individual delivering is the named, penalizable party (license suspension, personal fines), the role acquires a thin but real shield that a curbside robot cannot hold.
State-level rules requiring a licensed, medically-certified human operator physically aboard any commercial delivery vehicle — the pattern in California's AV framework (which still excludes vehicles over 10,001 lbs from driverless testing) and in bills like CA AB 316 (vetoed 2023) and comparable Teamsters-backed human-operator bills in Indiana, Illinois and New York. If such a rule passes in populous states and extends to sub-CDL vans, the seat becomes legally mandated rather than economically optional.
If autonomous middle-mile hauling matures first, the surviving human role becomes the exception-handler: refused deliveries, damaged freight adjudication, unsafe access decisions, recipient disputes, load securement calls. Task-mix shift toward the exception tier raises accountability, but only where employers designate the driver as the decision-maker rather than routing every exception to a remote dispatcher — which current telematics and camera-monitoring practice pushes the other way.
Scope creep at the doorstep — deliveries that include unboxing, appliance hookup, meter or lockbox access, signature plus ID capture, and returns pickup. Each addition puts more of the day inside unpredictable built environments (gates, elevators, stairwells, occupied homes) rather than at a curb, which is the portion no last-mile autonomy program currently addresses.
The limit. The structural ceiling is low. No license gates most of these jobs, pay is set by parcel volume, and the buyer is usually the shipper (Amazon, a retailer), not the resident who sees the driver — so there is no one positioned to pay a premium for a specific human. Monitoring technology is currently reducing judgment_accountability faster than task-mix shift raises it. Realistic upside is roughly 46 to the high 50s, and it concentrates in the specialty-courier and white-glove segments, not in high-volume parcel.
| New York-Newark-Jersey City, NY-NJ | 58,140 | $47,440 +6% |
| Los Angeles-Long Beach-Anaheim, CA | 40,900 | $46,500 +4% |
| Chicago-Naperville-Elgin, IL-IN | 34,680 | $47,950 +7% |
| Dallas-Fort Worth-Arlington, TX | 25,890 | $44,050 -2% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 20,930 | $45,740 +2% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 20,000 | $42,750 -5% |
| Houston-Pasadena-The Woodlands, TX | 19,050 | $43,920 -2% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 16,390 | $46,360 +3% |
| Anchorage, AK | 1,400 | $57,250 +28% |
| Hagerstown-Martinsburg, MD-WV | 1,610 | $53,190 +19% |
| San Jose-Sunnyvale-Santa Clara, CA | 4,480 | $52,100 +16% |
We have no reported case of a named organisation automating this occupation. Not one deployment, not one announcement.
That is worth saying out loud next to a score of 46. The verdict above is about what the work exposes — what current AI could do to these tasks. It is not a claim that anyone has done it. For this occupation those two things have come apart completely: the capability argument is on this page, and the evidence column is empty.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.