Artificial intelligence is saving clinicians an average of 132 hours each year — the equivalent of more than three full working weeks — yet a chronic lack of training is preventing the technology from realising its full potential within the health service, according to a major global survey and separate UK research.
Time savings and capacity gains
The Philips Future Health Index 2026, which polled more than 2,000 healthcare professionals and 20,000 patients across 10 countries, found that nearly half (46%) of clinicians reported annual time savings averaging 132 hours. Half of those surveyed said the technology had increased their capacity to see patients, allowing them to treat an average of eight more people per week.
In the UK, a separate study estimated that AI-powered administrative support could save NHS staff an average of 43 minutes per day — amounting to five weeks per year. A full rollout of such tools across the health service could free up as many as 400,000 hours per month, according to a pilot of Microsoft 365 Copilot across 90 NHS organisations. NHS England is now making the AI assistant available to more than 500,000 staff to help draft documents and analyse data more efficiently.
The administrative benefits of AI are already apparent: clinicians describe the technology as a “buddy” for discussing ideas, transcribing clinical notes and scheduling appointments. But its capabilities extend well beyond back-office tasks.
Clinical applications and diagnostic support
AI is being deployed to flag dangerous drug combinations, suggest diagnoses based on symptoms and assist in analysing X‑rays and medical scans. Clinicians have praised the technology for enhancing precision, aiding research and enabling detailed case analysis. The Philips Future Health Index has noted that planned investments in AI over the next three years show the biggest increase in critical decision support, with cardiology and radiology leaders showing particular interest.
In the NHS, AI is already used to help diagnose COVID‑19 from chest imaging and for secondary care dermatology referrals. A £21 million programme launched by NHS England in 2023 aimed to introduce AI for the diagnosis of chest conditions across 66 hospital trusts. Despite these advances, the technology’s deeper integration into clinical workflows remains patchy and inconsistent.
The training deficit undermining adoption
While nearly two-thirds (65%) of clinicians reported increasing their use of AI tools provided at work, a stark paradox has emerged: almost eight in ten (79%) healthcare professionals believe training for AI-enabled tools is limited or inconsistent at their organisation. The Philips survey found that 70% of clinicians described training as unavailable, limited or inconsistent — a figure that aligns closely with the broader global finding.
A 2026 survey of UK doctors conducted by the Royal College of Physicians (RCP) reinforced this picture: 79% said they needed training in clinical AI tools, yet 66% had no access to such support. The RCP survey also revealed that 68% of physicians believe the NHS lacks the digital infrastructure to introduce AI effectively, with 70% citing the inability to integrate AI tools with systems such as the Electronic Patient Record as the leading barrier.
The failure to provide adequate training and approved tools is driving clinicians to work around the system. The Philips index found that 64% of clinicians resort to personal AI tools when workplace options fall short. More than half (56%) of doctors are turning to personal AI systems because workplace offerings do not meet their needs, and a separate 2026 survey of UK physicians found that 69% use personal access to ChatGPT and Microsoft Copilot for clinical questions.
Shez Partovi, chief innovation officer at Philips, said: “The organisations aren’t moving fast enough to provide the tools and the training.” The report itself recommends that “expanding structured, role-specific training will help clinicians develop the digital skills and clinical judgment needed to work effectively with AI.”
Implementation of AI in NHS hospitals has proved more complex than anticipated. A UCL-led study published in The Lancet eClinicalMedicine in September 2025 identified key challenges including engaging clinical staff with high workloads, integrating new technology with aging IT systems, a general lack of understanding and scepticism among staff. Contracting alone took between four and ten months longer than anticipated, and by June 2025 a third of hospital trusts were not yet using AI diagnostic tools in clinical practice.
The RCP survey found that 70% of physicians support AI implementation in the NHS, but many remain sceptical about the health system’s digital infrastructure. Concerns over data privacy, bias, transparency and accountability persist. The UK is developing a regulatory framework for AI in healthcare, with the MHRA extending existing software regulations to encompass “AI as a Medical Device”. A National Commission into the Regulation of AI in Healthcare has been established to provide recommendations to the MHRA in 2026, with key themes including transparency, explainability, bias mitigation and cybersecurity.
Human oversight remains essential
Despite the promise of AI, the overwhelming consensus among clinicians is that it must remain under human control. The Philips survey found that 90% of professionals emphasised the need to maintain human involvement as AI advances, and 86% insisted that all AI outputs require human oversight.
The RCP survey underlined the grounds for caution: 73% of physicians are concerned about the risk of error, 54% about liability risks, and 52% about the potential for de‑skilling clinicians. As AI becomes more embedded in clinical workflows, the requirement for rigorous training and robust human oversight is seen not as an optional extra but as a fundamental prerequisite for safe and effective care.