45% of Hospitals Stuck in AI Pilot Phase, Blocked by EHR Vendors
Nearly half of large U.S. health systems cannot move beyond testing artificial intelligence tools, according to a survey of senior healthcare IT leaders. Electronic health record vendor dependencies and fragmented third-party software integrations are the primary culprits.
The Qventus survey, released Thursday, interviewed more than 60 chief information officers, chief AI officers, and other senior IT leaders at medium and large national health systems. Only 4% reported they had scaled AI with measurable outcomes.
The Execution Gap
Seventy-four percent of respondents cited EHR vendor dependency as a top barrier to AI deployment. Health systems wait for vendors to release AI features on their own timelines, then struggle to integrate multiple point solutions from different vendors without creating new bottlenecks.
A quarter of respondents lacked a clear process for measuring AI performance. Forty-two percent said they were actively deploying AI across multiple use cases, but scaling remained elusive.
Healthcare leaders are shifting how they measure success. Sixty-two percent now focus on revenue generation and 59% on hard dollar cost savings. Half see potential in autonomous platforms that handle scheduling, patient flow, and care gap identification with minimal human oversight.
The Cost of Waiting
Ninety-four percent of surveyed leaders said delaying AI deployment creates competitive disadvantage. Sixty-eight percent predicted further clinician burnout without AI tools to reduce administrative load.
The pressure is mounting. The population aged 65 and older will increase 42% by 2050, while healthcare faces federal spending cuts and workforce shortages.
Patient Trust Questions
A separate survey of 1,000 clinicians found most trust evidence-based AI tools and use them multiple times per patient encounter. Eighty-nine percent of clinicians said AI-driven clinical decision support leads to better outcomes.
Consumers tell a different story. Sixty-four percent of 1,000 surveyed U.S. health consumers said they would prefer to see a clinician who does not use AI at all.
How to Move Forward
Sutter Health demonstrated one path: establishing three guardrails, validating efficacy with real-world data from their own patient population, and using common AI infrastructure across the enterprise. The health system moved beyond pilots to system-wide implementations.
According to Sutter's imaging service line chair, integrating 10 to 20 separate point solution vendors into a health system cannot be done safely or at scale.
The stakes are high. A chief transformation officer at HonorHealth noted that healthcare margins are thin enough that betting wrong on technology can eliminate profitability entirely.
Learn more about AI for Healthcare and AI Agents & Automation to understand how these tools work in practice.
Your membership also unlocks:
