Most health systems that have proven AI can deliver value are failing to scale it, according to a new survey from Carta Healthcare that maps where hospital AI adoption breaks down.
The survey found 71% of organizations reporting measurable value from AI are not expanding those initiatives at pace. Carta CEO Brent Dover said the finding rules out doubt about the technology itself, and that what stalls is everything that comes after the proof.
EHR integration difficulty was the leading adoption barrier, cited by 44% of respondents, well ahead of clinician trust and regulatory concerns at 26% each. Dover said a tool can be accurate in a demo and still fail the moment a clinician has to leave their workflow to use it, and that added steps are where adoption quietly dies.
Clinical leaders now most frequently own AI strategy, surpassing both IT and executive leadership, yet 26% of organizations still have no clearly defined AI owner. Dover said a pilot without an owner has nowhere to go once the initial enthusiasm fades.
Vendor evaluation showed the strongest consensus, with 92% of respondents saying deep clinical domain expertise is critical when assessing an AI vendor. Dover advised health systems to ask vendors to prove integration in their own environment before signing, and to favor partners who share performance risk and can show peer outcomes.
The results suggest healthcare is entering a new phase where questions about whether AI can produce value give way to questions of governance, workflow and operational sustainability.
