AI tools are being built to summarize patient records and support clinical decisions, which leads many people to wonder whether doctors will be needed in the future. The data on jobs says they will. Medicine is among the professions least exposed to generative AI, because so much of it depends on examining patients, making accountable decisions and talking to people.
The short answer: on current evidence, AI is not on course to replace doctors. The ILO's 2025 index classes generalist and specialist medical practitioners, including specialties such as radiology, as not exposed to generative AI. US projections expect physician employment to keep growing. AI is more likely to change paperwork and decision support than the role of the doctor, and it raises the stakes on checking AI output. Forecasts can change as the technology develops.
How exposed medical jobs are
These figures are from the ILO's refined global index of generative AI exposure, which scores tasks from 0 to 1 and sorts occupations into bands from not exposed up to gradient 4, the most exposed.
| Occupation (ISCO-08) | ILO category | Mean task score (0 to 1) |
|---|---|---|
| Medical Secretaries | Gradient 3 | 0.53 |
| Pharmacists | Minimal exposure | 0.33 |
| Generalist Medical Practitioners | Not exposed | 0.29 |
| Specialist Medical Practitioners | Not exposed | 0.27 |
| Medical Imaging and Therapeutic Equipment Technicians | Not exposed | 0.22 |
| Dentists | Not exposed | 0.15 |
Administrative roles around medicine, such as medical secretaries, are far more exposed than clinical ones. Specialist medical practitioners, the ISCO-08 group that includes radiologists and other specialists, average 0.27.
Hiring outlook to 2035
The US Bureau of Labor Statistics projects employment of physicians and surgeons to grow 4 percent from 2025 to 2035, about as fast as the average for all occupations, with about 22,100 openings a year on average (BLS Occupational Outlook Handbook). The WEF's 2025 survey lists care economy jobs among those expected to grow significantly by 2030 (WEF Future of Jobs Report 2025).
What AI tools handle now
- Drafting clinical notes, letters and discharge summaries.
- Summarizing long patient histories.
- Surfacing relevant guidance and literature.
- Coding, billing and administrative paperwork.
- Decision support that flags possible findings for a clinician to review.
What stays with the doctor
- Examination and diagnosis: putting together what the patient says, shows and needs.
- Accountable decisions: treatment choices made and owned by a licensed person.
- Conversation: explaining options, risks and bad news.
- Procedures: the hands-on work of medicine and surgery.
- Checking the AI: NIST uses healthcare as its example of why confabulation matters: a confabulated summary of patient information could lead doctors to the wrong diagnosis or treatment (NIST AI 600-1). It also warns about automation bias, the tendency to defer too much to automated systems.
The job from here
For doctors, the most immediate change is in documentation and administration, which AI can draft and the clinician reviews. Decision support tools add a second opinion that still needs a doctor's judgment. The core of the job, examining, deciding and caring for patients, is where the ILO finds the lowest exposure.
Related guides: will AI replace nurses?, checking AI output and jobs least exposed to AI.
Pay and how to get in
The US Bureau of Labor Statistics puts the median annual pay for physicians and surgeons at $275,930 in May 2025, against $50,980 for all workers. The typical education needed to enter is: doctoral or professional degree. See how this job compares in AI exposure by pay and AI exposure by education level.
Questions and answers
Will AI replace doctors?
Current evidence says no. The ILO classes generalist and specialist medical practitioners as not exposed to generative AI, and the BLS projects 4 percent US growth for physicians and surgeons from 2025 to 2035.
Will AI replace radiologists?
Radiologists fall under specialist medical practitioners, which the ILO classes as not exposed to generative AI, with a mean task score of 0.27. AI tools may support image reading, but diagnosis and accountability stay with doctors.
Which medical jobs are most exposed to AI?
Administrative roles. Medical secretaries are in the ILO's gradient 3, with a mean task score of 0.53, far above clinical roles.
How will AI change doctors' work?
Mostly through documentation, summaries and decision support. These outputs need careful review, since NIST warns that confabulated patient summaries could lead to wrong diagnoses.
Can AI make medical mistakes?
Yes. Generative AI can produce confident but false content, and people can over-rely on automated output. That is why clinical review of AI output is essential.