Healthcare AI shifts from admin tasks to care transformation, virtual care growth stalls on finances

Healthcare organizations spend on AI at two to three times the rate of other industries, according to reporting by Healthcare IT News from the HIMSS AI in Healthcare Forum in Boston. The spending is real. The results, experts at the forum argued, are not yet matching the investment.

The core problem, as speakers including leaders from UMass Memorial Health and Stanford Healthcare described it, is that most providers are pointing AI at the wrong targets. Administrative tasks, scheduling, prior authorization, clinical documentation, are absorbing the bulk of AI capacity. Those are legitimate efficiency gains, but they fall short of what the technology can do for patient outcomes.

The automation trap

The distinction between automating a task and transforming a process matters more than it might seem. Automating a prior authorization workflow saves staff time. Redesigning care pathways with AI-assisted decision support can change whether a patient deteriorates before a clinician intervenes. Experts at the HIMSS forum said health systems need to pursue the second category with the same urgency they have brought to the first.

That shift is harder than it sounds. Clinical AI requires clinician buy-in, and trust remains a significant obstacle. In a separate Healthcare IT News interview, Jay Anders of Medicomp Systems addressed why many clinicians remain skeptical of AI tools. The concern is not simply that AI might be wrong. It is that clinicians often cannot see why AI reached a particular conclusion, making it difficult to know when to trust the output and when to override it.

Transparency in AI reasoning is becoming a practical requirement, not just a philosophical preference, as health systems try to move AI closer to the point of care.

Pilots that don't scale

A related pattern is showing up across health systems globally. Eric Wong, chief digital health officer at NHG Health in Singapore, told Healthcare IT News that organizations frequently launch AI pilots without first identifying the specific problems those pilots are meant to solve. The result is a graveyard of proofs of concept that demonstrate technical feasibility but never reach production.

Wong's framing points to a discipline problem as much as a technology one. The health systems making AI work at scale tend to start with the clinical or operational question and work backward to the tool, rather than starting with a new technology and searching for a use case to justify it.

Virtual care grows, but finances lag

AI adoption is unfolding against a broader digital health backdrop that has its own unresolved tensions. Virtual care utilization is rising in 2026, but Healthcare IT News reported that many health systems are losing money on their digital service lines. More patients are using telehealth and remote monitoring. Fewer health systems are breaking even on those services.

The gap between utilization growth and financial sustainability reflects structural issues that technology alone cannot fix. Reimbursement rates for virtual services in many markets still do not reflect the true cost of running a digital care operation. Infrastructure, platform licensing, care coordination, and clinician time add up faster than fee schedules have adjusted.

For health system executives, that means virtual care strategy in 2026 is as much a financial engineering problem as a technology deployment one. Systems that have found a path to profitability have typically integrated virtual care tightly with in-person workflows, using digital touchpoints to reduce avoidable high-cost encounters rather than treating telehealth as a standalone revenue line.

AI in revenue cycle: a clearer near-term win

One area where AI is demonstrating concrete financial results is revenue cycle management. Healthcare IT News profiled First Choice Neurology, where Dr. Ernesto Alonso described AI reducing cognitive burden for clinical staff while accelerating collections. The connection between documentation quality, coding accuracy, and revenue capture is direct enough that AI improvements show up in financial performance relatively quickly.

That makes revenue cycle one of the more mature AI use cases in healthcare, even if it falls squarely in the automation category that experts say should not be the end goal. For many organizations, it is the entry point that builds internal confidence and technical infrastructure for more ambitious clinical applications.

What comes next

The HIMSS AI in Healthcare Forum conversations and the virtual care financial data together point toward a consolidation phase in healthcare's digital transformation. The organizations that deployed broadly during the past several years are now under pressure to show that their investments produce outcomes, not just activity metrics.

On the AI side, that pressure is pushing the conversation toward governance, clinician adoption, and measurable clinical impact. On the virtual care side, it is pushing toward integration and reimbursement strategy. The CDC is also moving in parallel, with Matthew Ritchey of its Office of Public Health Data, Surveillance and Technology outlining plans for a secure public health data ecosystem designed to give clinicians and local agencies more timely, actionable information. That infrastructure, if it develops as described, would give health system AI tools better data to work with at the population level.