AI underused where it could deliver productivity gains, says RCR

The NHS is failing to capitalise on some of AI’s biggest productivity benefits by underusing the technology for administrative tasks, despite growing adoption across radiology and oncology services, new workforce census data from the Royal College of Radiologists (RCR) reveals.

Analysis from the RCR 2025 workforce census, seen by Digital Health News, shows that while AI is increasingly being adopted in clinical workflows, its use for routine administrative work remains limited, even in areas where departments report the greatest workload reductions.

Three-quarters (75%) of UK radiology departments now use AI, up from 69% in 2024, with the technology most commonly deployed to help clinicians identify abnormalities on medical images.

However, only 11% of departments use AI to draft reports, despite almost one in four (24%) of departments using it for this purpose reporting measurable workload reductions – the highest reported productivity benefit of any radiology AI application.

Administrative adoption remains low across the board. Just 13% of departments use AI for staff rostering, while 12% use it to manage referrals and appointments.

Dr Stephen Harden, president of the RCR, said: “Our findings make clear that AI has significant potential to help doctors manage rising demand, but its implementation in diagnostic and cancer services must be led by evidence.

“The NHS has begun introducing AI tools for administrative tasks, but these findings suggest there is considerable scope to expand their use where they can have a positive impact on productivity. AI does not mean we need fewer doctors.

“In fact, the UK is short of 2,300 clinical radiologists and 230 clinical oncologists needed just to meet current demand, with shortages set to worsen amid soaring demand for scans and cancer care.

“AI will deliver its greatest benefits for patients when we have the workforce, infrastructure and clinical oversight needed to implement it safely and effectively.”

The RCR’s annual report, published on 18 June 2026, is based on responses heads of service at all UK cancer centres and clinical directors of radiology departments.

The data also shows that more than four in five (81%) cancer centres now use AI for at least one purpose, compared with 63% in 2024. During 2025, two-thirds of centres introduced new AI tools, with 84% reporting a positive impact.

AI is now used by 81% of cancer centres to outline tumours and healthy organs before treatment planning, with clinicians reviewing the results. Among centres using the technology, 76% reported reduced workloads.

But data suggests AI remains underused for administrative tasks. Only 26% of cancer centres use AI to draft reports, despite the 29% of those doing so saying it cuts workload.

The RCR argues that wider implementation of proven administrative AI tools could help relieve mounting workforce pressures, allowing clinicians to spend more time delivering patient care rather than completing paperwork.

It is calling on the NHS to ensure all cancer centres have access to AI-enabled radiotherapy planning, expand the deployment of effective administrative AI tools across diagnostic services, and provide clinicians with practical guidance on selecting, implementing and monitoring AI safely.

The college said robust regulation, alongside adequate staffing and digital infrastructure, will be essential if AI is to deliver lasting productivity improvements without compromising patient safety.