What’s next for AI in surgery? Key takeaways from the Oxford Open Grand Round

AI is already changing healthcare - but not in the way many expect

While headlines often focus on AI performing surgery or replacing clinicians, the panel argued that the biggest impact today lies elsewhere. Administrative tasks such as clinical documentation, letter writing, appointment management and meeting notes are already being transformed by AI tools, reducing clinician workload and freeing up more time for patient care.

Rather than replacing surgeons, today's AI is helping to reduce cognitive burden and improve efficiency across the patient pathway.

Understanding AI beyond the buzzwords

The discussion clarified the differences between artificial intelligence, machine learning, deep learning and generative AI. Although public awareness has exploded since the arrival of tools such as ChatGPT, the technology itself has been developing over several decades. What has changed is accessibility - powerful AI models are now available to anyone with an internet connection.

From administration to the operating theatre

The panel highlighted opportunities for AI throughout the surgical pathway:

• AI-powered documentation and ambient clinical scribes.
• Automated pre- and post-operative patient communication.
• Decision support for clinicians.
• Image analysis and diagnostic assistance.
• Surgical navigation and future robotic applications.
• Training and personalised feedback for surgeons using video analysis.

While fully autonomous robotic surgery remains some way off, AI is already enhancing surgical planning, imaging and workflow, with significant advances expected over the coming decade.

Evidence must keep pace with innovation

A recurring theme was the need for robust evaluation. Hospitals are increasingly piloting AI solutions, but successful adoption depends on demonstrating measurable benefits - whether reducing administrative burden, improving patient outcomes or delivering financial value.

The speakers warned against adopting technology simply because it is innovative, stressing that implementation should always be supported by high-quality research and health economic evidence.

AI should augment—not simply automate

Drawing on experience developing AI tools within the NHS, Katrina emphasised that successful innovation starts by solving real clinical problems rather than deploying technology for its own sake.

Many current solutions still benefit from 'human-in-the-loop' models, where clinicians oversee or validate AI outputs. This hybrid approach balances efficiency with patient safety while the technology continues to mature.

Ethical challenges remain

The discussion explored several important questions:

• How should AI tools be validated before widespread clinical use?
• How do we address bias in training data?
• What happens when AI produces convincing but incorrect information?
• Could increasing reliance on AI lead to loss of clinical skills?
• How should patients be involved in AI implementation?

The panel agreed that patient safety must remain paramount, while recognising that AI should be judged against the realities of current healthcare rather than an unrealistic expectation of perfection.

Preparing the next generation of surgeons

One of the strongest messages was that future clinicians will need AI literacy alongside traditional clinical skills. Tomorrow's surgeons must understand not only how to use AI tools effectively, but also their limitations, risks and governance. 

The speakers argued that the greatest opportunity lies in developing clinicians who can bridge medicine, innovation and technology - helping to safely translate new advances into everyday clinical practice.

Looking ahead

The panel concluded that AI is unlikely to replace surgeons outright. Instead, surgeons who understand and effectively use AI are likely to outperform those who do not.

Ultimately, AI offers the potential to improve access to surgery, reduce administrative burden, enhance decision-making and improve patient care worldwide. Achieving that future, however, will depend on careful evaluation, strong governance and clinicians remaining actively involved in shaping how these technologies are developed and implemented.

Watch the full discussion on the NDS YouTube channel.