← Risk register SOC 53-3054 · reviewed 2026-08-11

Taxi Drivers

41,050 US workers · median $42,100/yr · Transportation

COOKED

The core task — driving a passenger from A to B, plus fare calculation, routing, and dispatch response — is the single task the AV industry has spent $100B+ to automate, and Waymo already runs paid driverless rides at scale in Phoenix, SF, LA, and Austin. Embodiment normally protects, but here the robot is the product: a car on mapped urban roads is a far easier physical problem than a crawl space or an ICU. What survives near-term is the human-contact residue: wheelchair and stretcher assistance, luggage, unaccompanied minors, medical transport, and airport/hotel runs where a named driver with a phone number is the service.

10-year outlook: Driverless fleets will take most dense-urban ride volume within a decade; the taxi jobs left will be assistance-heavy medical, paratransit, and rural work, at materially lower headcount.

Score — 32/100 resistance

Five dimensions, 0–20 each, summed. Higher means more protected. The arithmetic is shown so you can check it: 6 + 11 + 5 + 6 + 4 = 32.

Task resistance 6/20

Core tasks are already automatable. Fare metering, dispatch acceptance, GPS routing and lane-keeping are all shipped features in a Waymo or Zoox vehicle today, which puts this at 6 rather than 0 only because curbside pickup in double-parked traffic, loading a folded wheelchair into a trunk, and dealing with a passenger who vomits or won't get out still need hands.

Embodiment 11/20

Some physical or field component. You are in a vehicle on public roads all shift — lifting suitcases, assisting elderly riders in and out, occasionally shoveling snow off the hood — but the road itself is mapped, signed, and geometrically constrained, which is why this sits at 11 instead of the 16+ of a lineman on a pole or a plumber under a house.

Liability shield 5/20

Certification preferred, not legally required. A chauffeur endorsement or TLC hack licence is a city permit with a background check and a medical, not a professional licence that makes you personally answerable for an outcome; the fleet's commercial policy and the TNC's insurance absorb the claim, so at 5 the paperwork barely slows a driverless competitor down.

Trust premium 6/20

Some relationship component. Street hails and app dispatch are anonymous by design and the passenger is rating a 5-minute interaction, which caps this at 6 — the points come from the minority of the trade with a repeat book: the hotel doorman's regular, the standing dialysis run, the school contract where the parent wants the same face every morning.

Judgment & accountability 4/20

Executes defined procedures on defined inputs. Route choice, when to refuse an intoxicated fare, and how to handle a no-show are governed by company policy, meter rules, and municipal regs, and a supervisor or the app resolves disputes — the discretion is real but small-stakes and reversible, hence 4.

Scored twice. An independent second run returned 31/100 — COOKED, agreeing with the verdict above.

This score sits on a verdict boundary. At 32/100 it is one point from EXPOSED. Re-scoring moves results by a point or two, so here the score is more informative than the label.

Confidence: high · reviewed 2026-08-11 · how scoring works

Tasks already automatable

What survives

Active moats: embodiment, trust

How to future-proof this job

Field report — do you do this job?

Has AI actually changed your work?

Self-reported and unverified — a sentiment signal, not a survey. One response per person per occupation; you can change your answer.

From people who do this job

Nobody has filed one yet. If you do this work, you know things the rubric can't see.

What has actually changed in your work?

Concrete beats general: a tool that arrived, a task that moved, a headcount decision you watched happen. Don't include anything that identifies you or your employer if that would put you at risk.