WVU panel: operations must own AI

HealthSystemCIO reported on August 5 that executives from WVU Medicine, Ballad Health and St. James’s Hospital Dublin said health systems are struggling to make AI education stick when information technology teams lead adoption on their own. Ilo Romero, vice president and associate CIO at WVU Medicine, Pam Austin, former senior vice president and CIO at Ballad Health, and Guido Giunti, chief data officer at St. James’s, argued that operational and clinical leaders have to own the workflow changes AI is meant to support. The panel’s point was narrower than a general call for “AI readiness.” Their argument was that hospitals often treat AI as a software rollout when the harder work is assigning business ownership, setting measures and deciding how exceptions get handled in live workflows such as prior authorization, patient access and revenue cycle management. That framing came through in HealthSystemCIO’s summary of the discussion, which said AI literacy fails when IT drives adoption without operational ownership. (healthsystemcio.com) ### Why did these leaders say IT cannot carry AI adoption by itself? Pam Austin said AI literacy initiatives struggle when IT tries to drive adoption without operational and clinical leadership, according to HealthSystemCIO’s coverage and a related video clip. The reason, the panelists said, is that trust sits closer to the frontline manager, department head or clinical leader who owns the work, not with the technology function alone. (healthsystemcio.com) WVU Medicine’s Romero and St. James’s Giunti made the same point from different angles. HealthSystemCIO said the panelists argued that operational and clinical leaders “hold the trust adoption depends on,” a line that places ownership with the people responsible for day-to-day performance, not just the team implementing the tool. ### What does “operational ownership” look like in practice? (healthsystemcio.com) Prior authorization, patient access and revenue cycle management are the kinds of functions the panel was talking about. In those areas, AI tools may summarize documents, draft responses or sort work, but the business unit still has to decide who owns the queue, what counts as a successful outcome, when a human intervenes and how unresolved cases move to the next team. Those requirements follow directly from the panel’s emphasis on operational ownership and measurable outcomes. (healthsystemcio.com) Defined exception paths are central to that model. If an authorization request is incomplete, if a payer response conflicts with clinical documentation, or if a patient-access task stalls between systems, the panel’s logic is that an operations leader must own the handoff and the resolution rules. Without that structure, AI may speed up one task while leaving staff to manage the gaps around it. That is an inference from the panel’s stated focus on workflow ownership, governance and measurement. (healthsystemcio.com) ### Why does the panel tie literacy to measurable outcomes? HealthSystemCIO’s August 5 article described a “role-based, federated model” for building workforce capability that outlasts any single tool. That suggests the panel was not arguing for one-time training sessions or generic AI awareness. They were describing literacy as something attached to a person’s actual role, the process they own and the outcomes they are expected to improve. (healthsystemcio.com) Guido Giunti has made that case before. In a March 3 HealthSystemCIO interview, he said St. James’s Hospital Dublin was building its digital transformation around a literacy-first model, using short sessions, peer champions and formal courses tied to immediate use cases. He also said many staff mistakenly think AI adoption is just another software deployment. (healthsystemcio.com) ### What is the practical message for revenue-cycle teams? WVU Medicine, Ballad Health and St. James’s were describing a management problem as much as a technology problem. In frontline revenue-cycle teams, an AI project will need a named business owner, operating measures and governance that includes the people running the process if leaders want adoption to last. That is the thread connecting literacy, workflow ownership and performance measurement in the panel discussion. (healthsystemcio.com) The August 5 HealthSystemCIO article and podcast are the next place to look for the full discussion, including comments from Romero, Austin and Giunti on how health systems assign ownership and train staff around AI-enabled workflows. (healthsystemcio.com)