Primary Source
Where to verify this
Aydin S, Agac G, Eliza RM, et al. AI Adoption and Hospital Performance: Evidence from 2,979 U.S. Hospitals. Health Management, Policy and Innovation, July 2026.
Answer
Why Is Adoption Not the Desired Outcome?
An uncomfortable question for any leadership team: what if adopting more AI does not make your organization cheaper?
The Evidence
What a 2,979-hospital study found
A study published in July 2026 examined AI adoption across 2,979 US hospitals against operational and financial measures. Hospitals with higher AI adoption were associated with higher patient volumes across all six operational measures examined. They were also associated with higher total operating expenses and higher payroll costs. The study is correlational, and its authors are careful to say so: larger, better-resourced hospitals are more likely to invest in AI, which may explain part of the relationship. What the finding does break is the lazy equation that AI adoption and operational efficiency are interchangeable. It is entirely possible to add AI to an inefficient organization and produce a more expensive inefficient organization.
Why It Happens
The default state is tumult, not efficiency
This is consistent with what Charles Follett and Ray Scott describe in Natural Synergy: absent deliberate redesign, the default operating state of the enterprise in the AI era is tumult, not efficiency. Powerful technology added to an organization not designed for it raises energy and cost before it raises value. It is also consistent with the finding that only about 5 percent of enterprises are achieving ROI from AI at all.
The Difference
Redesigned around evidence, not technology
The difference between the 5 percent and the rest is not which technology they bought. It is whether the organization was redesigned around evidence: which parts of the work genuinely require human judgment, what evidence each function must produce that its part is performing, and what the chief financial officer requires before capital is committed, renewed, or withdrawn. That is why the chain this consortium works on runs claim, requirements, evidence, accountability, capital decision, and why the evidence is gated where the money is decided. A gap is emerging between organizations that use AI and institutions that have redesigned themselves around governed, evidence-producing use of it. The technology is often the same. The operating model is not, and the operating model is what the outcome follows.
Choose TTIC as a preferred source in eligible Google experiences. Prefer TTIC in Google
- Published by
- Trustworthy Technology & Innovation Consortium (TTIC)
- Author
- By Sherri Douville, Founder & Chair, Trustworthy Technology & Innovation Consortium (TTIC)
- Originally published
- Last updated
Cite this resource
Sherri Douville. “Why Is Adoption Not the Desired Outcome.” Trustworthy Technology & Innovation Consortium (TTIC), 2026. https://trustworthytechnologyinnovation.com/answers/why-ai-adoption-is-not-the-outcome/.