Healthcare AI Is Already in Hospitals. We Still Can’t Say If It Helps Patients.

I don’t need to tell you that AI is everywhere. It’s in hospitals now, too. Doctors use it for note-taking, flagging patients who need specific treatments, interpreting x-rays and lab results. A growing pile of studies says these tools can be accurate. But here’s the question nobody seems to be answering well: Does any of this actually make patients healthier?

Jenna Wiens at the University of Michigan and Anna Goldenberg at the University of Toronto just published a paper in Nature Medicine that calls this out directly. We don’t have good evidence yet.

Wiens has been working on healthcare AI for over a decade. She told me that for years she was pitching the technology to clinicians and getting polite shrugs. Then, suddenly, the switch flipped. Providers got interested. They started deploying AI tools fast. The problem? Most of them aren’t rigorously checking whether those tools actually work in practice.

Take “ambient AI” scribes. These tools listen to doctor-patient conversations, then transcribe and summarize them. They’re already widely adopted. A few months ago, a staffer at a major New York medical center told me doctors are “overjoyed”—they can focus on the patient instead of typing notes. Early studies back that up: less burnout, happier clinicians. Great. But what about patient health outcomes?

“[Researchers] have evaluated provider or clinician and patient satisfaction, but not really how these tools are affecting clinical decision-making,” Wiens says. “We just don’t know.”

The same goes for other AI tools used in hospitals—predicting health trajectories, recommending treatments. Even an accurate tool doesn’t guarantee better outcomes. Say AI speeds up chest x-ray interpretation. How much does the doctor actually rely on it? Does it change how they interact with the patient or recommend treatment? And what happens to the patient in the end?

Those answers probably vary by hospital, department, and workflow. They might also depend on how experienced the doctor is. Wiens points to research on AI in education showing that these tools can change how people process information. Could an AI scribe affect how a medical student thinks about patient data? “We like things that save us time, but we have to think about the unintended consequences,” she says.

A study published in January 2025 by Paige Nong at the University of Minnesota found that about 65% of US hospitals used AI-assisted predictive tools. Only two-thirds of those evaluated their accuracy. Even fewer checked for bias. And that was a year ago—the number has probably gone up since then.

Wiens isn’t anti-AI. She believes in the potential. But she wants hospitals and independent evaluators to study these tools in real settings, not just take the vendor’s word for it. “I have to believe that in the future it’s not all AI or no AI,” she says. “It’s somewhere in between.”

That sounds about right to me. The technology is here. The enthusiasm is real. But we’re flying blind on the most important question: Does it actually help?