More than 230 million people turn to ChatGPT to ask questions related to health and wellness each week. Meanwhile, four out of five physicians are exploring the use of AI in their practices, and the large language model Claude for Healthcare, rolled out in January, is poised to transform medical billing.
As new capabilities for artificial intelligence (AI) in healthcare emerge almost daily, healthcare leaders face a paradox: urgency to adopt alongside persistent hesitation to scale.
There are also signs that AI could improve access to health information and guidance for vulnerable people, with 600,000 health-related queries to ChatGPT coming from underserved rural areas each week.
Yet surging capital costs continue to erode the money available for strategic investment. As a result, AI use cases in healthcare often struggle to move past the pilot stage. In this era, making the right bets on AI in healthcare depends on responsible AI strategies as well as a willingness to act quickly to seize opportunities at scale.
Go beyond AI experimentation
Adoption of AI in healthcare more than doubled the rate of adoption in other industries from 2023 to 2025.
Optimizing AI’s business value in healthcare depends not just on deploying the right tools in the right areas, but also on scaling these tools quickly. It also necessitates careful consideration around how to move fast with AI, without putting undue stress on an already-challenged workforce.
For example, while agentic AI shows tremendous promise for reducing administrative burden and costs in the healthcare revenue cycle, across industries there are barriers to its adoption. One of the biggest roadblocks is the fear of being replaced by AI: a concern that two out of three leaders surveyed say is fueling hesitancy in agentic AI adoption.
And while use cases for AI are expanding in healthcare, an analysis by KLAS Research indicates most organizations use AI for lower-risk purposes like ambient AI to generate structured notes from physician-to-patient conversations, imaging triage, predictive risk modeling and patient message response.
Moving the needle on AI ROI
With limited resources to invest in AI, how can healthcare leaders make the right moves for near-term and long-term value? Here are three considerations in a transformative environment.
Don’t just taste the tool. Chase the tools. Capital constraints demand a shift from experimentation to implementation. Rather than testing isolated tools, organizations should prioritize solutions designed to scale from the outset. For healthcare leaders, this means looking for tools that may have been built for one workflow or hospital but could transfer seamlessly to others. This approach accelerates deployment, reduces development costs and strengthens foundations for data governance and workforce adoption.
In determining which AI initiatives to pursue, it’s also important to map their value to the healthcare organization’s strategy. This helps avoid isolated AI use cases that struggle to deliver on their intended value. It also ensures that the investments an organization makes around AI are thoughtfully aligned with the health system’s mission, vision and business objectives, and carefully balance risk with rewards.
Ensure that the tools you choose are compatible with the EHR. This budget-protecting approach paves the way for rapid-paced innovation through seamless integration. It also makes it easier to engage team members in an AI tool through ease of access and integration with other functions. The ability to draw from data in the EHR also provides a foundation for actionable intelligence based on data that is clean, governed and accessible. With this foundation, healthcare teams can more easily turn pain points into viable innovations that present strong potential for ROI.
In 2026, key opportunities to link AI innovation to integration in the EHR include:
- Clinical workflow improvement, including documentation, imaging and diagnostics
- Patient engagement, such as by enhancing patient interactions through the use of virtual assistants, AI-powered symptom checkers and educational resources
- Revenue cycle management, where AI can convert unstructured data into usable information and streamline prior authorization, billing, coding and denials
- Controls to monitor AI value. A recent EY global risk study found that organizations with real-time performance monitoring for AI initiatives are 65% more likely to achieve cost savings (Editor's note: This article's author is affiliated with EY). This underscores the importance of embedding AI properly within the wider digital ecosystem and regularly tracking AI efforts with support from an AI governance committee. The committee should focus on value metrics, such as operational efficiency, quality improvement, cost reduction and patient outcomes, to ensure AI delivers measurable benefits and aligns with organizational goals.
As the pace of AI adoption in healthcare increases, developing a roadmap for implementation that emphasizes rapid deployment and reward while reducing risk will drive momentum. It will also build team members’ confidence in innovation—critical for sustaining AI gains over the long term.
John Ward is a partner in technology consulting at EY and is head of health technology.