Editor’s note: This article incorporates key insights from Healthcare Dive’s recent live event, “AI in Healthcare.” You can watch the full event here.
Four years after the rapid emergence of generative artificial intelligence tools in the clinical landscape, the healthcare sector finds itself at a critical inflection point. While the initial wave of excitement centered on the potential to alleviate clinician burnout, optimize administrative workflows, and streamline complex billing processes, the honeymoon phase has ended. Today, payers and providers are confronting the stark realities of implementation: prohibitive costs, the persistent "black box" of algorithmic decision-making, and an evolving, often chaotic regulatory environment.
As the industry matures, the conversation has shifted from "can we use AI?" to "how do we use AI safely, equitably, and profitably?" During the Healthcare Dive “AI in Healthcare” event held on November 4, industry leaders outlined a roadmap for the future, emphasizing that successful integration requires more than just high-quality software—it requires a fundamental shift in institutional culture and governance.
The Chronology of Adoption: From Hype to Hesitation
The trajectory of AI in healthcare has been anything but linear.
- 2022–2023: The Era of Speculation. Following the explosion of large language models, hospitals rushed to pilot AI tools for documentation and note-taking. The primary driver was the desperate need to address widespread clinician burnout.
- 2023–2024: The Reality Check. As tools moved from pilot programs to enterprise-wide integration, the challenges of interoperability and data security surfaced. Concerns regarding the ethical implications of AI errors—specifically those that could lead to clinical misdiagnoses or patient harm—began to dominate board-level discussions.
- 2025–2026: The Focus on ROI. Current efforts are concentrated on “agentic AI”—systems capable of autonomous task execution. Organizations are now pivoting away from “innovation for innovation’s sake” toward high-impact, revenue-positive applications, such as revenue cycle management and automated prior authorizations.
1. Bridging the Trust Gap: The Clinician’s Perspective
The most significant barrier to the widespread adoption of AI remains human, not technical. Clinician buy-in is the litmus test for any new tool, and currently, the results are mixed. According to a 2023 American Medical Association survey, 41% of 1,000 physicians reported feeling “equally excited and concerned” about healthcare AI. This ambivalence stems from a fear that algorithms may replace human judgment or, worse, introduce errors that clinicians are held liable for.
Micki Tripathi, chief artificial intelligence implementation officer at the Mayo Clinic, argues that the secret to overcoming this skepticism lies in "clinician-led co-design."
"AI tools are accepted the most among clinicians when they’re developed by clinicians themselves," Tripathi explained during the event. He cited a standout success at the Mayo Clinic: a tool designed to support nurse-to-nurse patient handoff summaries during shift changes. Because the nurses who would actually use the tool were involved in the conceptualization and development phase, they became the internal champions for the technology.
However, Tripathi acknowledges that this model is difficult to scale, particularly in highly specialized fields. For a cardiologist working in a hyper-specialized sub-department, a generic AI model often lacks the nuanced clinical expertise required for high-stakes decision-making. These gaps in expertise remain the primary friction point for adoption.
2. Strategic Investment: Prioritizing "Agentic" AI in Cash-Strapped Systems
For rural and under-resourced hospitals, the promise of "agentic AI"—technology that can perform autonomous actions like scheduling, data entry, and communication—is tempered by severe budget constraints.
Dr. Alexander Fortenko, vice president of innovation at Ochsner Health, advises that hospitals with limited capital should prioritize administrative bottlenecks that directly impact patient and provider satisfaction. Prior authorization, for instance, remains one of the most inefficient aspects of the US healthcare system. By automating the communication between providers and insurers, hospitals can reduce administrative overhead and accelerate patient care.
Lydia Daniels, VP of analytics and AI at Epic, emphasizes that in a climate of financial instability, tools that improve revenue cycle management are the only ones that guarantee a clear return on investment (ROI). "Coders have the capacity to do more, to take on additional specialties," Daniels said. "You’re not necessarily lowering your headcount, but you’re doing more with what you have. And in rural medicine, that is the only strategy you can have."
3. The Financial Paradox: Is AI Saving Money or Driving Costs?
A central tension in the current healthcare landscape is the debate over AI’s economic impact. While the goal of AI is efficiency, there is growing concern that it may inadvertently inflate costs. Some payers have warned that AI-driven documentation tools are enabling providers to “upcode” more aggressively, leading to higher billing for the same level of care.
This has resulted in an adversarial “arms race.” Payers are deploying their own AI to audit claims, while providers are using AI to ensure their documentation maximizes reimbursement. Shri Santhanam, chief product officer at Machinify, suggests that this tension may actually be a necessary catalyst.
"Does this crisis almost work to our benefit in that it forces a dialog, which would otherwise have been lurking in the background?" Santhanam posited. By bringing these inefficiencies to the surface, the adversarial dynamic may force the industry to finally address systemic billing issues that have persisted for decades.
4. Governance in the "Wild West": Navigating the Regulatory Patchwork
The absence of a comprehensive federal regulatory framework for AI has left health systems to navigate a fragmented landscape of state laws. David Arnold, senior vice president and chief AI and data officer at Hackensack Meridian Health, described the current situation as the "wild west."
"There is no single comprehensive regulatory framework to date from the federal government," Arnold noted. "But things are coming fast and wild on the state side."
For large health systems operating across state lines, this creates a logistical nightmare. Compliance strategies that work in one jurisdiction may be insufficient or illegal in another. Aashima Gupta, global director of healthcare at Google Cloud, warns that without federal preemption, AI governance within large organizations risks becoming “unsustainable.”
In response to this vacuum, industry-led coalitions have emerged to set standards. Lucy Orr-Ewing, director of policy and strategy at the Coalition for Health AI (CHAI), advocates for a focus on AI literacy. CHAI emphasizes that governance is not merely a legal or IT function; it is a clinical and cultural one. By training nurses, physicians, cybersecurity experts, and data scientists on a "multidisciplinary front," organizations can ensure that errors are caught early, rather than discovered after they impact patient outcomes.
Implications: Building a Sustainable Future
As the industry moves toward 2027, the focus must shift from the novelty of AI to the sustainability of its integration. The implications for the healthcare sector are clear:
- Shift to Outcomes-Based Governance: Hospitals must move beyond surface-level policies and embed AI oversight into the clinical workflow. This includes ongoing audits of algorithmic performance to ensure equity and accuracy.
- The Rise of Human-in-the-Loop Systems: The most successful AI models will be those that function as "co-pilots," augmenting human expertise rather than attempting to replace it.
- Collaborative Ecosystems: The adversarial relationship between payers and providers over AI-driven claims must evolve into a cooperative framework. Shared data standards and transparent AI utilization could save billions in administrative waste.
- Literacy as Infrastructure: Just as hospitals require electricity and water, they now require a workforce that is AI-literate. Investments in training will be as vital as investments in hardware.
Ultimately, the successful adoption of AI will not be measured by the sophistication of the algorithms, but by the ability of healthcare leaders to foster trust among their clinicians and transparency for their patients. The technology is no longer the bottleneck; the human and structural elements of the healthcare system are. Solving for these will define the next decade of medical innovation.
