SAFE
The job is physically positioning frightened, often frail patients to sub-millimeter tolerance on a linear accelerator, verifying setup against imaging, and running the beam — none of which current robotics can do in an uncontrolled clinical room. AI is already absorbing the adjacent screen work (treatment plan checks, auto-contouring, image registration, dose record documentation, scheduling), so the informatics side of the role thins while the daily hands-on delivery persists. Patients see the same therapist every weekday for six weeks, which makes the relationship, not the button-press, the durable part.
Roughly flat across the period, with year-to-year wobble.
Median pay $85,560 → $105,310 -1.5% 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
+1.9% 19,200 → 19,600 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +1.9% more of these jobs by 2034. Note that safe does not mean well paid — several of the fastest-growing resistant occupations are among the lowest paid on the register.
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.
~900 openings a year on average, including replacing people who leave.
DosimetristRadiology TherapistRadiologic TherapistRadiation Therapist (RT)Staff Radiation TherapistRadiation Therapy TechnicianRegistered Radiation TherapistRadiation Therapy Technologist (RTT)Radiation Oncology Registered Nurse (Radiation Oncology RN)Computed Tomography Simulation Therapist (CT Simulation Therapist)
Holding it up: embodiment . Weakest point: judgment & accountability .
Mixed — a routine tier and a judgment tier At 13 the manual setup — immobilization masks, tattoo/surface-mark alignment, couch shifts from CBCT match, and daily side-effect checks — stays human, but a real share of the shift (record-and-verify checks, IGRT auto-registration, chart QA, ARIA/Mosaiq documentation, appointment sequencing) is already being handed to software, which is what keeps it out of the 14+ band.
Hands-on in uncontrolled environments 17 reflects that the work happens with hands on a semi-clothed, sometimes nauseated or intubated patient in a vault, lifting and rolling them onto an index bar, seating a thermoplastic mask, and clearing a 6 MV gantry's collision path — variable bodies and variable rooms, though the machine and the vault itself are fixed and controlled, which is why it isn't 19.
Licensed human required and personally liable 13 fits ARRT(T) certification plus state licensure in most states, with the therapist personally signing the daily treatment record and the wrong-site/wrong-dose misadministration reportable under state radiation control regs — but the prescription, dose, and plan approval belong to the radiation oncologist and medical physicist, so the deepest liability sits above you.
Meaningful discretion 10 is right because the prescription, fractionation, and tolerances are written for you and out-of-tolerance shifts get escalated to physics or the physician, but you decide in real time whether today's match is good enough to treat, whether a shift needs a re-sim, and whether to stop the beam — genuine calls made alone in the console, inside a defined envelope.
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 (13/20 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 (13/20) is whether the law requires a licensed human to sign. Trust premium (14/20) is whether buyers specifically pay for a person. Judgment and accountability (10/20) 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 37 of this occupation's 67 points (55%).
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.
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 79/100, still SAFE.
NRC/state radiation-control 'medical event' reporting rules or a Joint Commission element of performance naming a credentialed therapist as the required human check on AI-generated online adaptive replans (as on Varian Ethos / Elekta Unity workflows), so an autonomous adaptive plan cannot be delivered unsigned
Task-mix shift as auto-contouring, plan QA and dose documentation are absorbed: the residual day becomes adaptive-therapy decision support, immobilization problem-solving on frail/pediatric/claustrophobic patients, and toxicity triage — the tier machines can't reach. Genuine two-tier occupation, and the routine tier is the one already going
Formal adoption of online adaptive radiotherapy at scale, where the therapist at console owns the accept/reject call on a daily AI-recontoured plan under time pressure with the patient on the table — ASTRO/AAPM guidance (e.g. AAPM TG-report work on adaptive QA) codifying that authority rather than routing it to a physicist
State radiologic technologist licensure laws (e.g. the long-pending federal CARE Act, plus states like Idaho/Missouri that still lack RT licensure) extending mandatory licensure to all radiation therapists, combined with ARRT/ASRT-backed rules requiring a licensed therapist to personally attest to daily setup verification and any AI-suggested shift before beam-on — making the attestation individually, not institutionally, liable
Limited route. Patients do not select their therapist and payment is bundled per-fraction under Medicare's radiation oncology payment structures, so there is no consumer-side price signal for a human. The only realistic lift is proton/pediatric/SBRT centers marketing named continuity-of-care teams, which is marginal
The limit. Embodiment is near its practical maximum at 17 and cannot rise much; the honest risk is not displacement but headcount compression per linac as adaptive automation lets one therapist cover more fractions, which the register's dimensions don't capture.
| New York-Newark-Jersey City, NY-NJ | 960 | $135,550 +29% |
| Los Angeles-Long Beach-Anaheim, CA | 480 | $169,150 +61% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 390 | $119,960 +14% |
| Charlotte-Concord-Gastonia, NC-SC | 360 | $99,250 -6% |
| Houston-Pasadena-The Woodlands, TX | 350 | $105,770 +0% |
| Phoenix-Mesa-Chandler, AZ | 350 | $108,770 +3% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 310 | $92,060 -13% |
| Atlanta-Sandy Springs-Roswell, GA | 280 | $99,860 -5% |
| San Jose-Sunnyvale-Santa Clara, CA | 130 | $206,350 +96% |
| San Francisco-Oakland-Fremont, CA | 220 | $205,670 +95% |
| Sacramento-Roseville-Folsom, CA | 110 | $182,240 +73% |
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 67. 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.