Since the onset of the AI chatbot boom, tech companies, insurance industry mouthpieces, and healthcare administrators have spun a yarn that introducing AI to healthcare is a surefire way to lower costs for patients.

Mario Schlosser, co-founder and chief technical officer of the medical insurer Oscar Health, has stated that AI was the “only way” to reduce the cost of seeing a doctor in the US over the next three to five years. In 2024, a report by the analyst firm McKinsey argued that AI could generate healthcare savings “as high as $360 billion annually.”

Not only have those bold pronouncements fallen flat, but it sounds like the exact opposite is true. In reality, both hospital administrators and health insurers agree that AI tools have caused health care costs to rise, according to healthcare industry publication Stat.

The main issue revolves around AI “scribes,” which are tools that transcribe live doctor-patient conversations into clinical notes. These scribes, healthcare insiders told the publication, have not translated into financial savings more broadly.

“Right now, ambient scribes are inflationary, and that’s a problem,” Caroline Pearson, executive director at Peterson Health Technology Institute told Stat. “We need technology to help us lower health care costs.”

There are three main causes. The first has to do with the notes themselves. Burnt-out doctors used to take shortcuts on paperwork, writing only the bare-bones notes before sending them off to billings. That meant they were getting paid for “simple” visits, even for cases that are actually complicated. Now that AI is around to record every last detail, many visits are being rated at a higher complexity level, which is used to justify higher billing rates.

The second has to do with software frequently “nudging” doctors to add diagnoses they discussed but didn’t jot down. “Now you have a tool that’s saying, ‘hey, you talking about their UTI, but you didn’t add it to your visit diagnosis. Do you want to add it now?'” chief medical information officer of FMOL Health Bobby DuPre told Stat.

Third is straight up volume — when doctors aren’t stuck typing notes, they have time for more visits. At DuPre’s FMOL, for example, clinicians using AI scribes saw 22 percent more patients overall. As the executive puts it, the “more people they see, the more payment they get.”

All of this has compounded into a system where the tools that were supposed to make healthcare more affordable have actually had the opposite effect — a potent reminder that no technology can magically solve the tension between healthcare driven by profit, and the basic human need for affordable care.

More on AI in healthcare: America’s Largest Hospital System Ready to Start Replacing Radiologists With AI, Its CEO Says