RADIOLOGISTS have called for renewed strategies to tackle the workforce crisis, amid growing concerns that AI radiology tools might be failing to streamline workload and address clinician burnout.1
Radiology and the Workforce Crisis
There is a 30% shortfall of clinical radiology consultants in the UK,2 predicted to increase to 39% by 2029 without further action, according to data from The Royal College of Radiologists.3
Speaking on an expert panel at the Health + AI Tech Show on 29th April 2026, Azeem Alam, specialist registrar in clinical radiology, Imperial College London, London, UK, said: “The median age for a consultant radiologist to leave the NHS is 50-years-old and that’s now dropping to 45-years-old in the next few years.
“We’re almost 2,000 consultant radiologists short and 500,000 scans a year are not reported within 28 days.
“So, there is a huge amount of burnout within radiology.
“There’s around a one in 10 competition ratio to get into radiology and attrition at the end of that funnel – so there are massive problems within radiology and the workforce crisis right now.”
House on Fire: Will AI Increase Workload in Radiology?
In 2024, more than half of Clinical Directors in radiology found the use of AI tools made no significant change to their workloads.3
In fact, 37% reported an increase in workload directly resulting from the implementation of AI tools.3
Alam continued: “A lot of my experience looking at AI tools in radiology feels like they’re trying to install smart lighting in a house which is already on fire.”
AI tools praised for high sensitivity are reportedly often built from pre-training data on specific demographics, model settings, and standards, neglecting to account for local differences in workflow.
Speaking on an AI tool for lung nodule detection on CT, Alam said: “What we find is that it may detect one true lung nodule, but it will also detect nine false nodules and what our clinical director wants us to do, as radiologists, is to explain why each of those false positives are, indeed, false.
“So, you can see the increase in workload that now takes and what we end up doing is ignoring the tools all together.”
He added that 35% of AI-assisted radiology tools wind up unused 18 months after the contract is signed.
Radiologists Versus AI: Who is Liable for Error?
In a study published last month, researchers presented mock jurors with a hypothetical malpractice scenario, in which a patient suffered irreversible brain damage because a radiologist failed to detect a brain bleed from a CT scan.3
Mock jurors were almost 50% more likely to side against the radiologist when the clinician only reviewed the CT once after it was flagged by AI, compared with when the radiologist read the scan both before and after receiving feedback from AI,3 contributing to concerns surrounding workload increase.
Whilst panellists unanimously agreed that ultimate liability falls with the responsible radiologist, Alam said: “The liability often doesn’t take into account all the systemic and workflow issues which have precipitated those mistakes in the first place.”
Experts called for the development of AI tools within existing platforms and realistic contexts in the NHS, rather than in a vacuum free from the systemic issues facing radiology, including a mounting need to grow the workforce.
References
1 Alam A et al. Imaging reimagined. Can radiology stay in control of AI? Panel discussion. Health + AI Tech Show, 29 April, 2026.
2 The Royal College of Radiologists. 2023 clinical radiology and clinical oncology workforce census reports. 2024. Available at: https://www.rcr.ac.uk/news-policy/latest-updates/2023-clinical-radiology-and-clinical-oncology-workforce-census-reports/#:~:text=These%20delays%20are%20the%20direct,workforce%20only%20grew%20by%203.5%25. Last accessed: 29 April 2026.
3 The Royal College of Radiologists. Clinical radiology: workforce census 2024. 2025. Available at: https://www.rcr.ac.uk/media/4imb5jge/_rcr-2024-clinical-radiology-workforce-census-report.pdf. Last accessed: 30 April 2026.
4 Bernstein et al. The radiologist-AI workflow and the risk of medical malpractice claims. Nat Health. 2026;1:386–389.
Featured image: ATRPhoto on Adobe Stock
