
Will AI Replace Radiologists?
AI is already finalizing some normal chest X-rays and acting as a second reader in breast screening. We found no verified radiologist layoffs, and BLS still projects job growth.
One chest X-ray product listed on Sectra's Amplifier Marketplace states plainly: "Confirmed normal studies are reported and signed off automatically, without a radiologist read." Software, not a radiologist, finalizes those reports.
For radiologists, the impact is narrow but real. AI is taking over specific, repeatable reading tasks: clearing normal chest films, acting as the second reader in breast screening, and flagging urgent CT findings. Abnormal studies, complex cases and legal accountability for the final call still belong to people.
Where AI already reads scans in radiology departments
These deployments are documented in our evidence. Each one replaces or speeds up a defined slice of the reading workload. None replaces the whole job.
- Autonomous normal chest X-rays. Oxipit says it launched ChestLink in April 2022 as the first CE Class IIb-certified autonomous AI application for healthy chest X-ray reporting. In a study summarized by Oxipit, ChestLink identified normal studies with 99.8% sensitivity and could remove 36.4% of normal chest X-rays from a primary care dataset.
- Breast screening second reads in Sweden. Stockholm-region hospitals plan to use Lunit as an independent second reader for 200,000–250,000 annual breast examinations, after routine use at an initial hospital. The Discoveries in Health Policy blog summed it up: "AI replaces a defined second-reading task in Sweden."
- Screening trials. In the MASAI trial, run within the Swedish national screening program, ScreenPoint Medical reports that AI-supported reading improved detection and cut radiologist workload. The Lancet paper found a 12% reduction in interval cancer rate. A UK trial that replaced the second human reader in NHS screening found equivalent performance and 46% less human reading.
- CT triage at scale. Aidoc says its technology runs in nearly 2,000 hospitals and analyzes approximately 60 million patient cases annually, and it is supplying infrastructure to a consortium of 12 hospital systems. In January 2026 it secured clearance for what was described as the first multi-condition AI triage solution for body CT.
- Broad uptake in the UK. The Royal College of Radiologists' 2025 census found 75% of radiology departments now use AI.
Executives are also talking openly about replacement. The CEO of America's largest public hospital system, Katz, said: "We could replace a great deal of radiologists with AI at this moment, if we are ready to do the regulatory challenge." He asked fellow CEOs why they should not push for changes to New York rules. That is a stated intention, not a job cut.
We found no reliably sourced layoffs of radiologists attributed to AI.
Skills and specialties gaining value as AI spreads
The evidence points to work that sits around the algorithm rather than competing with it.
- AI governance and local validation. In May 2026 the ACR and SIIM approved the first practice parameter for imaging AI, which asks facilities to set up AI governance. That role needs radiologists who understand both reads and models.
- Catching drift. One review reports that models can lose up to 24% specificity outside their training institution. Someone has to notice that before patients pay for it.
- Clinical context. An experiment with radiologists found that contextual information improved performance while AI predictions did not, on average. Pairing history and prior studies with the image is a human edge.
- Accountable sign-off. In both UK and Swedish screening trials, a human radiologist remained accountable throughout. The multi-society RSNA joint statement also guides purchasers and users on monitoring AI tools.
Field note: Most UK departments already use AI, but the RCR census shows hospitals deploy it mainly for clinical tasks and underuse it for admin tasks that cut workload. If you want relief from the backlog, push for admin tools first.
Myths vs reality for radiologists
| Myth | Reality |
|---|---|
| AI already reads most scans on its own. | xAID reports that essentially all clinically deployed radiology AI is designed to assist a radiologist. Normal chest X-ray sign-off is a narrow exception. |
| Handing radiologists AI makes everyone more accurate. | Researchers found radiologists "underweight AI predictions", so the gains did not show up on average. |
| Vendors stand behind the diagnosis. | An AuntMinnie analysis notes "vendors explicitly disclaim diagnostic responsibility". |
| BLS projects 9% radiologist growth. | One site cites 9%, but the current BLS projection is +3.4% for 2025-35. |
What the numbers say about radiologist jobs
The BLS projections count 28,700 radiologist jobs in 2025, rising to 29,600 by 2035. That is +3.4%, almost the 3.5% average for all occupations, with about 700 openings a year and a median wage of $420,860.
BLS has already built AI into its thinking. Its Monthly Labor Review notes that "radiologists are incorporating AI into traditional computer-aided diagnosis (CAD) systems, which increases efficiency."
Exposure scores, all via secondary aggregators, put radiology in the middle of the pack. One reports that Anthropic ranks radiologists 120th of 756 for observed AI exposure. Another cites an ILO exposure of 45%, and a third reports 50% overall and 76% theoretical exposure. High exposure on paper has not yet translated into fewer jobs.
Our verdict: AI takes the routine reads, not the radiologist
We think AI will keep peeling off the most repeatable work: normal chest films, screening second reads and urgent-case sorting. Hospital leaders like Katz want more, but liability, regulation and real-world accuracy gaps stand in the way today. Radiologists who run, validate and overrule these tools look well placed.
Questions people ask
Will AI replace radiologists?
Not as a whole profession, in our read. BLS projects radiologist jobs rising from 28,700 in 2025 to 29,600 in 2035, while AI takes narrow tasks such as normal chest X-ray sign-off and screening second reads.
Can AI read X-rays without a radiologist?
For a narrow slice, yes. Oxipit's ChestLink is listed as signing off confirmed normal chest X-rays automatically without a radiologist read, but xAID reports that essentially all deployed radiology AI is designed to assist a radiologist.
Is radiology still a good career choice?
BLS projects +3.4% growth for 2025-35, about 700 openings a year and a $420,860 median wage. The work is shifting toward overseeing AI and handling harder cases.
Who is liable if AI misses a finding?
An AuntMinnie analysis notes that current AI is approved only for clinical decision support and that vendors explicitly disclaim diagnostic responsibility. In the UK and Swedish screening trials, a human radiologist remained accountable throughout.
Sources
Every number, quote and example on this page was checked by script against the sources below before publishing: quotes must match the source word for word, and figures must appear in a source or in BLS data. Our verdict is editorial judgment. Spotted an error? Tell us.
- U.S. Bureau of Labor Statistics: Employment Projections, Radiologists (29-1224)
- AI Medical Imaging in 2026: Best Radiology AI Tools, FDA Clearances, and Diagnostic Accuracy | Pinggy Blog (pinggy.io)
- 12 leading hospital systems team to advance radiology AI (radiologybusiness.com)
- Discoveries in Health Policy: Using Radiology Business "AI" Section to Understand Trends in 2025-2026 (discoveriesinhealthpolicy.com)
- FDA-Cleared AI Radiology Tools Hospitals Actually Use (buildmvpfast.com)
- CEO of America’s largest public hospital system says he’s ready to replace radiologists with AI (radiologybusiness.com)
- AI in Radiology: The Complete 2026 Guide | xAID (xaid.ai)
- Replacing radiologists doesn’t remove risk; it moves it to patients. (auntminnie.com)
- AI in Radiology: Why Demand for Humans is Growing 9% - UDS (udshealth.com)
- RSNA Publishes Joint Statement on Use of AI Tools in Radiology (rsna.org)
- Interval cancer, sensitivity, and specificity comparing AI-supported mammography screening with standard double reading without AI in the MASAI study: a randomised, controlled, non-inferiority, single (sciencedirect.com)
- ScreenPoint Medical: AI-Enhanced Mammography Screening Significantly Improves Breast Cancer Detection and Reduces Radiologist Workload - ScreenPoint Medical (screenpoint-medical.com)
- Three Futures for the Diagnostic Radiologist: A Structured Disagreement About What AI Actually Changes (arxiv.org)
- CXR Suite | Oxipit (oxipit.ai)
- Oxipit CXR Autonomous AI - Sectra Amplifier Marketplace (amplifiermarketplace.sectra.com)
- ChestLink Autonomous AI Application Clinical Performance (oxipit.ai)
- Will AI Replace Radiologists? A Decade of Evidence (2026 Data) (aichanging.work)
- Combining human expertise with artificial intelligence : experimental evidence from radiology - Catalog - UW-Madison Libraries (search.library.wisc.edu)
- Google Scholar (scholar.google.com)
- RCR Census: AI Underused for Admin Tasks That Cut Workload (auntminnieeurope.com)
- Industry and occupational employment projections overview and highlights, 2024–34 : Monthly Labor Review : U.S. Bureau of Labor Statistics (bls.gov)
- Will AI Replace Radiologists? Risk Score, Tasks, Growth (2026) (aijobimpactcalculator.com)
- Will AI Replace Radiology Techs, Sonographers and MRI Techs? 2026 Outlook and What to Do (onehour.digital)