AI in healthcare: Shifting the frame from ‘what’ to ‘how’

The conversation around Artificial Intelligence in healthcare is evolving. We are moving beyond the hype of algorithms and automation, and beginning to ask the more important questions: How do we ensure AI serves our people and patients? How do we adopt it in ways that strengthen the health and care system as a whole? What kind of leadership do we need to make that happen? More than that, how do we harness the reality of AI now even as we set a path for AI in the future? 

It is no longer enough to explore what AI can do. We must focus on how it should be done and why. As NHS leaders, we sit at a pivotal moment. We are facing relentless operational pressures but also possess a rare opportunity to shape the long-term future of care using assets that are available now and maturing at pace. The choices we make today will set the foundations for the NHS of tomorrow. 

This was powerfully reinforced during the recent UCLPartners and Health Foundation roundtable following their report AI in London Healthcare: The Reality Behind the Hype. The session, chaired by Caroline Clarke, Regional Director for NHS England London, brought together system leaders, clinicians, regulators, and innovators to explore how we scale AI responsibly, equitably, and effectively. The discussion was frank, ambitious, and future facing. 

The clear takeaway for me was that AI must be rooted in service transformation, not ad hoc pilots or disconnected initiatives. We need to stop seeing AI as a standalone product and instead view it as part of a broader shift in how we anchor innovation in service redesign, workforce development, and population health priorities. Less of a what and more of a how. This perspective generates ripples of different thinking in terms of how we sufficiently govern but suitably liberate the potential of AI. 

From a provider perspective, AI is already having an impact. We are seeing smarter triage, faster more accurate diagnostics, and improved targeting of care. In one example, UCLPartners’ collaboration with NHS North-East London and Health Navigator uses AI-driven coaching to reduce pressure on emergency services. However, these benefits come with new challenges. As we increase speed of diagnoses, how do we ensure we retain the humanity in diagnostic conversations? When we leave radiographers as opposed to doctors or nurses to deliver same-day CT results, they are often asked to hold sensitive patient conversations, these are conversations they were never trained to hold. This is not a technical gap. It is a leadership challenge. 

As NHS leaders, we must:  

• become fluent in the implications of AI across the system 

• understand how AI reshapes roles, impacts workflows, and redefines risk 

• be cognisant of the differential impacts of these aspects on patient, citizen and staff populations and 

• create space for innovation to thrive safely and sustainably, with the right governance and support in place. There is a need to ensure sufficient controls to ensure safety, but enough freedom to allow innovation.  

There is a question on whether we regulate a product or instead agree principles for ways of working that are agnostic of mushrooming products but, sufficiently detail the fundamental specifications of those products.   

Equally, we must shift the way we engage with the innovation. Too often, technological tools are in search of a hypothetical problem rather than real, articulated service needs. Our role is to clearly define the real-world challenges we need solved and invite innovation into that space. That clarity will ensure that AI development is purposeful, impactful, and aligned with patient and service priorities.  

Infrastructure is critical. The digital divide across systems and regions is still wide. Many of our best clinical innovators are pulled away due to a lack of protected time, investment, or leadership support. This is a loss we cannot afford. If we want AI to work at scale, we need to build the systems, talent, and culture to support it. 

London is well placed to lead this work. We have the diversity, the partnerships, and the ambition. But we must go further. We need stronger alignment across providers, clearer procurement and evaluation frameworks, and a collective voice to shape market development and regulatory policy. 

That is why I am encouraged by the approach UCLPartners is taking. They are not chasing the latest gadgets but focusing on solving real system problems using the right technology, applied in the right way. Their leadership is helping to reframe the conversation: from excitement about what AI can do to a serious planning for how we make it work. 

The future of AI in healthcare will not be decided by algorithms. It will be shaped by leadership, strategy, and our collective willingness to invest in the right foundations today. If we want AI to be a force for good, we need to lead with purpose, act with urgency, and plan for the long term. 

The time to build that future is now, with clarity, courage and foresight.