By Cathy Hoole, General Manager of Operations, PRO Radiology
Current evidence does not suggest that AI is replacing radiologists in Australia. Radiology training positions continue to grow internationally, demand for diagnostic imaging remains strong, and AI is increasingly being used to support radiologists in their work.
Cathy Hoole has spent more than a decade at PRO Radiology working alongside FRANZCR radiologists as they've built careers in teleradiology. Every time a new AI tool makes headlines, radiologists at different career stages ask some version of the same question: is this the beginning of the end for the profession?
It's a fair question. But a decade of real-world developments gives us a much clearer picture of where the profession is heading.
What the Radiology Workforce Evidence Actually Shows
Radiology workforce and training data show continued demand for radiologists, even as AI becomes more capable and widely adopted.
In 2016, Geoffrey Hinton, the Turing Award-winning computer scientist often called the "godfather of AI", famously suggested hospitals should stop training radiologists, predicting that deep learning would outperform them within five years.
It became one of the most repeated predictions in the AI-and-jobs debate. A decade later, AI has made substantial progress in medical imaging, but the predicted disappearance of radiologists has not occurred.
United States
- Diagnostic radiology residency programs offered 1,241 positions in 2026, up from 1,210 in 2025, according to the National Resident Matching Program.
- That's an increase of approximately 2.6%, with radiology programs continuing to expand training capacity.
- Across diagnostic and interventional radiology, 97.6% of available residency positions were filled in the 2026 Match, although applicant numbers have declined from their recent peak.
Sources: NRMP 2026 Main Residency Match and Radiology Business analysis of the 2026 results.
Australia
- Australia continues to face radiology workforce capacity pressures, with imaging demand placing sustained pressure on available reporting resources.
- The introduction of Australia's National Lung Cancer Screening Program in July 2025 has created an ongoing requirement for low-dose CT screening and specialist assessment.
- The Royal Australian and New Zealand College of Radiologists (RANZCR) advocates for the safe and responsible integration of AI into clinical practice, including human oversight of autonomous AI systems.
Sources: Australian Government – National Lung Cancer Screening Program and RANZCR – Artificial Intelligence.
The American figures provide useful evidence of training trends, although they should not be treated as a direct measure of Australia's radiology workforce.
The important distinction is that advances in AI capability have not translated into the wholesale workforce replacement predicted a decade ago.
What AI Can and Can't Do in Radiology Today
AI can assist with image analysis, prioritisation and defined diagnostic tasks, but clinical interpretation also depends on patient context, specialist judgement and professional accountability.
Some AI systems can identify findings on medical images, flag potentially urgent cases and assist radiologists in prioritising their workloads. These are meaningful advances, particularly when imaging services are dealing with high reporting volumes.
However, performance on a defined diagnostic task is different from performing the full role of a radiologist.
| Area | What AI does well | Where human expertise remains essential |
|---|---|---|
| Task | High-volume, well-defined image analysis and triage | Interpretation of ambiguous or atypical findings |
| Example | Identifying suspected abnormalities and prioritising cases | Correlating imaging with patient history, previous scans and the referrer's clinical question |
| Complex cases | Supporting detection and measurement of specific findings | Applying subspecialty judgement in complex MSK, neurological and oncological imaging |
| Accountability | Supporting workflow and decision-making | Taking responsibility for reports, communicating critical findings and discussing results with referring clinicians |
Why Benchmark Performance Isn't Clinical Performance
AI can perform exceptionally well on carefully selected datasets. But a model achieving high accuracy on a benchmark does not establish that it can independently manage the range of cases encountered in everyday clinical practice.
An incomplete patient history, an unusual presentation or a finding that contradicts previous imaging may require interpretation extending well beyond pattern recognition.
There's also an important limitation to acknowledge: some AI systems are already capable of performing increasingly complex diagnostic tasks, and autonomous interpretation of narrowly defined examinations is an emerging area of development.
That could change how some routine examinations are reported, the mix of work available to radiologists and potentially the economics of certain reporting services.
Australia's professional and regulatory framework remains important here. RANZCR advocates for a human-in-the-loop approach to autonomous AI, while the Therapeutic Goods Administration regulates AI-enabled software where it meets the definition of a medical device and is not excluded from regulation.
The technology will continue to improve. The question is how safely and effectively it can be incorporated into clinical practice.
Why Demand for Radiologists Remains Strong
Demand for radiology reporting is being supported by increasing imaging requirements, an ageing population and expanded screening programs, while AI offers opportunities to improve workflow efficiency.
Australia's imaging services already face pressure from growing demand, including the need to provide timely reporting to metropolitan, regional and rural healthcare providers.
The National Lung Cancer Screening Program is a practical example. Launched on 1 July 2025, it provides eligible high-risk Australians with low-dose CT screening intended to detect lung cancer earlier.
These examinations require appropriate interpretation, follow-up and clinical management.
PRO Radiology is directly involved in this work, partnering with Lung Screen Australia to provide specialist reporting of low-dose CT scans through the National Lung Cancer Screening Program.
AI may help imaging providers manage increasing volumes by supporting detection, prioritisation and workflow. However, it would be misleading to suggest AI is the primary reason imaging demand is growing.
Population ageing, clinical utilisation, screening and advances in imaging technology all contribute.
For radiologists, increased productivity through AI could help address reporting backlogs and capacity constraints. It does not automatically mean that every productivity improvement creates additional jobs, and the longer-term workforce effects remain uncertain.
That distinction matters when considering how radiology careers may change over the coming decade.
The Future of Radiology Jobs: What This Means for Your Career
AI is changing the tasks radiologists perform and the technology they use, making adaptability, subspecialty knowledge and clinical judgement increasingly important career skills.
As AI tools become more integrated into reporting workflows, radiologists may spend less time on certain repetitive tasks and more time reviewing complex findings, resolving uncertainty and applying specialist expertise.
This is often described as a human-in-the-loop model, where AI assists with defined tasks while the radiologist retains oversight and responsibility for the clinical report.
For those currently training in radiology, developments in AI are worth understanding and preparing for. They are not, on current evidence, a reason to assume the specialty has no future.
Why This Matters for Flexible and Remote Reporting
The work of reviewing scans, applying clinical judgement and communicating findings can increasingly be supported through secure remote reporting platforms.
That makes teleradiology an established option for FRANZCR radiologists who want greater control over their working arrangements without stepping away from clinical practice.
At PRO Radiology, we understand how important flexibility, reliable technology and administrative support are to the radiologists working with us.
The takeaway isn't to wait and see how AI develops. It's to become comfortable using evolving reporting technology while continuing to develop the clinical expertise that patients and referring doctors depend on.
Radiologists interested in greater flexibility can explore remote reporting careers with PRO Radiology.
Where PRO Radiology Fits Into This
PRO Radiology combines experienced FRANZCR radiologists, remote reporting technology and dedicated operational support to deliver diagnostic reporting services across Australia.
Established in 2009, PRO Radiology is an independently owned Australian teleradiology provider supporting imaging centres, healthcare providers and hospital services.
Technology has always been central to the way we operate. Our reporting workflows use InteleViewer, Fluency for Imaging voice recognition and embedded AI tools to support efficient remote reporting.
Our reporting platform and workflow are designed to help radiologists focus on clinical interpretation, backed by administrative and technical support.
That operational support matters. AI can help make reporting more efficient, but radiologists still need reliable systems, manageable workloads and access to clinical colleagues when difficult cases arise.
For FRANZCR radiologists considering the next stage of their careers, whether that's more flexibility, greater control over caseload or the opportunity to work with evolving reporting technology, PRO Radiology welcomes a confidential conversation.
Explore radiologist careers with PRO Radiology or contact our team to discuss your options.
You can also email recruitment@proxraymanagement.com.
