AI news ·
HonorHealth CTO warns health systems to limit AI spending until clinical outcomes are proven
HonorHealth's CTO urges systems to cap AI spending until they prove clinical outcomes. He cites ambient documentation and predictive models as the only tools delivering value.
Dr. James Whitfill, SVP and CTO of HonorHealth, warns health systems against rushing into AI investments without clear, proven outcomes. He will deliver the keynote "AI in Clinical Medicine: Are We There Yet?" at the HIMSS26 APAC Conference in Singapore, arguing that without structured spending limits, healthcare organisations risk draining budgets and upending clinical workflows.
The tension between promise and proof
Whitfill describes AI as "a seismic force" affecting every industry, but he sees healthcare caught between what generative AI can handle in single interactions and the hard evidence from peer-reviewed literature. "I believe health systems need to have real, but limited, investments in AI to see how it works in their environment before scaling more widely," he said. The reason for limits is straightforward: large-scale spending can easily collide with negative impacts on clinical workflow, and AI "does not always work as expected when deployed in a health system." Moving from a pilot to enterprise use, he insists, must follow a return on outcomes, efficiency, or experience proven inside the organisation itself.
Predictive AI: The overlooked workhorse
While industry chatter gravitates toward agentic AI, Whitfill points to predictive models as the real workhorse few notice. "Predictive AI remains the unsung hero of AI that few are paying attention to," he said. He cites ECGs, retinal imaging, and straightforward radiology studies as examples where simple tests fed into models can surface diagnostic insights that directly aid physicians. Throughput in hospital systems and operating room optimisation are also areas where real-world AI solutions already produce measurable gains.
Ambient documentation delivers, but the physician shortage persists
Ambient documentation tools have seen rapid uptake at HonorHealth with tangible effects. "Our physicians have qualitatively told us that as little as one week later, they could not imagine practising without it," Whitfill said. Data shows clinicians complete charts within the same day far more often with ambient support. Yet the fundamental physician shortage remains nearly as acute as ever. Even with these gains, he said, "we are not yet seeing where AI is fundamentally impacting this shortage." The organisations that pull ahead are the ones that force outcome measurement from every AI initiative, rather than assuming technology alone will close the gap.
Governance and literacy over magic
Organisations that treat AI not as magic but as a technology to scrutinise get better results, Whitfill said. That means building AI literacy across teams and establishing governance that oversees AI from intake through ongoing operations. When a health system releases an AI tool, it should understand exactly how data-including sensitive demographics-feeds predictions, test for bias, and watch for model drift over time. As health systems integrate AI for Healthcare, Whitfill's focus on governance and literacy separates those seeing consistent value from those simply accumulating tools.
Sorting hype from reality
Whitfill remains grounded about the pace of change. "Despite bold claims from a number of AI evangelists, we have yet to see dramatic changes in how we practice medicine," he said. Frontier models are developing fast, but he urges conference attendees to keep their eyes on peer-reviewed literature. "Remaining grounded by monitoring what works and what doesn't remains an important guide in sorting out hype from real impact."
Why this matters for healthcare
Health systems face enormous pressure to adopt AI, but Whitfill's message is clear: move deliberately, measure relentlessly, and never outsource judgment to a model. Start with small, bounded investments tied to specific outcomes. Build governance that tracks bias and drift. Invest in physician literacy so teams know what the technology can and cannot do. In a field where hype runs high and capital is finite, the organisations that prove value in their own four walls will be the ones that actually improve care.