■ MODERATE RISK ■ Healthcare
AI has already transformed the planning side of radiation therapy and automation is steadily absorbing setup verification — but a frightened patient on a treatment table, thirty sessions in a row, still needs a human running the room. The role evolves toward oversight; it doesn't disappear.
“AI targets tumors precisely. But positioning the patient still needs a human.”
Our AI replacement risk score — how we score jobs
Radiation therapists are the clinicians patients actually see through cancer treatment. The daily rhythm: verifying the prescription and plan, positioning patients on the linac table with millimeter precision using lasers, masks, and immobilization devices, acquiring cone-beam CT images to confirm alignment, delivering the fraction while monitoring from the console, and watching for skin reactions, weight loss, and the emotional wear of a person attending their twentieth of thirty sessions. It's equal parts technical operation and sustained human contact with people on the worst ride of their lives.
AI has already reshaped the upstream. Auto-contouring of organs and targets, AI-generated treatment plans, and automated plan QA — work that consumed physicist and dosimetrist hours — are in routine clinical use. On the treatment floor, surface-guided systems track patient position continuously, image-matching software aligns cone-beam CTs automatically, and adaptive-therapy platforms adjust plans to daily anatomy with algorithmic speed. Each advance trims manual steps from setup and verification, nudging the therapist role from operator toward supervisor of increasingly self-checking machinery.
But the physical and human core holds. Positioning a post-surgical patient with limited shoulder mobility, judging whether today's swelling warrants calling the physician, halting a beam because something looks wrong, and carrying a patient's morale across weeks of daily encounters — these need a trained human in the room, and regulation agrees: licensed therapists must deliver treatment. Cancer incidence rising with an aging population underwrites demand. Our risk score of 50 reflects productivity, not replacement — more patients per therapist as automation compresses setup time, slower staffing growth than the caseload alone would predict, and a role that keeps its license while shedding its manual steps.
Automatability: our editorial assessment of current and near-term AI capability
The planning side has already been transformed — auto-contouring and AI planning are standard of care in many centers now. Through 2030, automated setup verification and adaptive therapy keep compressing time per fraction, letting each therapist treat more patients and slowing staffing growth. Regulation, safety accountability, and rising cancer incidence keep licensed humans at the console through the 2030s and likely beyond; this is a role being streamlined, not staffed out.
Not the role, but it's already replaced chunks of the workflow — auto-contouring, AI treatment planning, and automated image matching are in clinical use. Delivering treatment still legally and practically requires licensed therapists who position patients, monitor delivery, and assess people daily. Expect more patients per therapist, not therapists per unemployment line.
Solid, with a caveat. Cancer caseloads are rising with the aging population, and licensed therapists remain mandatory. But automation is raising throughput per therapist, so job growth will lag caseload growth. Candidates comfortable with rapidly evolving technology — adaptive platforms, surface guidance — will find the strongest demand.
The patient-shaped parts. Physically positioning people with pain, anxiety, or limited mobility; judging when a skin reaction or weight change needs the physician; stopping treatment when something's off; and sustaining a person emotionally through weeks of daily fractions. Automation verifies coordinates — it doesn't hold a hand.
Chase the frontier equipment. Experience with adaptive radiotherapy, MRI-guided linacs, surface-guided setup, and proton systems makes you the therapist departments compete for, because those platforms need humans who understand both the automation and its failure modes. Add advanced certification or dosimetry training, and invest deliberately in patient-care excellence.