AI in Healthcare: From Promise to Practice

Healthcare has never had more technological promise or more pressure to deliver on it than it does today.

Innovations with technology are staggering. Generative AI is drafting appeals, summarizing clinical notes, powering ambient tools, and enabling at-home patient engagement. Over 96% of U.S. inpatient hospitals now use EHR systems. This should be the age of seamless, intelligent care. But somewhere between potential and practice, momentum gets lost.

Legacy infrastructure, fragmented governance, workforce fatigue, and widening resource gaps continue to slow progress. Even more challenging is the fact that payers, providers, and patients, are advancing at their own pace, each building digital capabilities without a common cadence.

Meanwhile, the pressure to provide better care with less is building. More than 700 U.S. hospitals, many in rural areas, are at risk of closure. Legislative changes could further shrink coverage for millions.

At this moment, it is not point solutions but scaled innovation that can truly transform care. To scale innovation sustainably, healthcare must embed it within real workflows, ground it in interoperability, govern it with intent, and build for alignment across the system.

Everyone’s innovating. So why does it still feel disconnected?

The problem begins when innovation happens in isolation. Health systems are experimenting with GenAI and digital tools, but without shared infrastructure or enterprise-wide alignment, these pilots rarely scale.

Only one in four systems has governance models in place to responsibly manage GenAI use, and most still wrestle with fragmented data environments. Instead of simplifying care, this often adds more complexity to the way clinicians work.

Take the revenue cycle, for instance, AI can now generate appeals in minutes, yet payers still process them manually. It creates asymmetry and drives up administrative costs.

What it takes to scale AI in healthcare

To move forward, leaders must design for convergence. It means making innovation a part of how care actually works: connecting the dots across teams and ensuring every effort brings in better outcomes for all key stakeholders.

Here’s what that shift looks like in action:

1.  Redesigning the workforce, not replacing it

Scalable innovation in healthcare begins with a hard truth: healthcare systems won’t move the needle unless they rethink how care teams actually work. In 2024, 57% of health system executives cite workforce shortages as a top strategic concern. Lack of workforce readiness is also among the top three roadblocks to digital transformation. This emphasizes a widespread gap between deployment and human readiness on the ground.

Forward-looking providers are responding in various ways:

• They’re investing in workforce resilience. Nurses are being upskilled for hybrid, tech-enabled roles, not to replace clinical intuition, but to strengthen it.
• They are deploying GenAI tools that reduce cognitive burden. For example, ambient documentation helps clinicians automate note-taking and flag readmission risks. Pre-visit summaries are also becoming essential, as they surface patient context ahead of appointments to streamline care delivery.
• And they’re reclaiming time and capacity by reimagining workflows. Workflow redesign, paired with smart delegation, has the potential to deliver 15-30% time savings per shift, enough to bridge a gap of nearly 300,000 inpatient nurses^[8].

These are enablers of a more sustainable care model. Innovation must be grounded in the experience of those who deliver care in order to succeed.

2.  Building change management frameworks for AI

There’s no one-size-fits-all approach for leveraging AI in healthcare. Because this is not just another technology rollout.

Unlike cloud migrations, where infrastructure leads, AI demands that we first understand the work, what requires cognition, what creates friction, and where support is most needed. Centers of Excellence help providers get this right.

These centers formalize governance, align workflows, and ensure safety, equity, and trust in deployment. Without them, innovation risks stalling at the surface, useful in theory, but detached from the practice of care.

At Johns Hopkins, a predictive bed management dashboard co-designed with frontline teams became an integral part of daily decision-making. That’s what integration looks like. For AI to scale, it must first fit into the rhythm of care.

3.  Bridging the trust gap in clinical AI

Innovation isn’t uniformly welcomed across the healthcare enterprise. AI has found its footing in healthcare’s back office, but in clinical settings, it’s still finding its voice. Automation is scaling fast where the stakes are lower, like billings and appeals, but when it comes to diagnosis, triage, or care planning, hesitation runs deeper. This is understandably so; frontline clinicians are asked to trust tools they didn’t help build, in environments where errors carry real human costs.

That doesn’t mean clinical innovation should come to a halt. It means it must be guided differently.

For AI to make a real difference in clinical practice, it has to ease the clinician’s workload. The opportunity lies in supporting clinicians with tasks like, population health risk stratification and surveillance, patient history summarization, and capacity management. When AI complements decision-making, reduces cognitive fatigue, and fits naturally into the way care is delivered, it builds trust.

4.  Redefining ROI beyond dollars

We need to view ROI from a broader perspective if we need to scale AI in healthcare. When we define ROI by cost savings and budget cuts, we may overlook what is truly important. Success should show better outcomes and a stronger connection between clinicians and patients.

In an environment where so much of the work that matters such as care coordination, clinical summarization and provider-patient engagement, isn’t directly billable, return on investment can’t be measured in dollars alone. It must account for time reclaimed, trust built, and care delivered more thoughtfully.

Forward-looking health systems are starting to shift the conversation. They’re focusing on what improves care rather than measuring success solely by what gets automated. Are we making everyday tasks easier for clinicians? Are we freeing up time to be present with patients? These are the questions that must be answered with clarity every day.

Reimagining healthcare AI through human-led care

The next frontier for healthcare AI is its augmentation. Systems are shifting from back-end automation to patient-facing intelligence, utilizing AI that helps book care, triage symptoms, and interpret longitudinal records to inform decisions. Designed right, these tools build trust, reduce cognitive burden, improve access, and free up time for patient connection.

Nearly 60% of healthcare CEOs now rank GenAI as a top investment priority, and 79% remain optimistic about long-term growth. Still, 70% cite regulatory uncertainty as a key barrier to scale.

The path forward demands bold provider leadership. Progress won’t come from flashy deployments or quick wins. It will come from doing the work that truly moves the system forward. It includes eliminating systemic waste, creating shared data foundations between payers and providers, putting in place a strong change management framework, and staying focused on measurable value, both financial and non-financial.

It’s time we start shaping AI into something more foundational, reliable, transparent and deeply attuned to the realities of care. The impact of AI lies in quietly and seamlessly enabling every workflow, every decision, every interaction. And in the end, the real progress is how meaningfully we bring technology closer to the people it’s meant to serve.