Everyone agrees AI scribes are increasing health care costs. No one agrees what to do about it

STAT News – Wednesday, April 8, 2026
By Brittany Trang

To hear health systems tell the story, artificial intelligence tools like ambient scribes are helping not only reduce doctor burnout, but also increasing payments from insurers that haven’t been compensating them properly. But on insurer earnings calls, the payers position themselves as white knights sounding the alarm on providers using AI to raise health care costs to an unsustainable level.

However, at least behind closed doors, both sides appear to agree that AI scribes are driving up health care costs.

“The investors, the health plans, and the providers, in private, were like, ‘OK, well, it’s quite clear scribes are increasing coding intensity. One hundred percent,’” said Caroline Pearson, executive director at the Peterson Health Technology Institute, describing a roundtable PHTI held earlier this year. The nonprofit institute, founded in 2023, evaluates the impact of new technologies on health care costs and quality.

The question that remains is what the health care industry is going to do about this technology-driven inflation, especially in the face of political pressure to lower the cost of health care. Health economists told STAT they predict the AI arms race between providers and payers will end in payers setting lower reimbursement rates, which will hit under-resourced providers the hardest.

Health insurers have three options, Pearson said: pay the increased costs, downgrade expensive visits to less-expensive tiers, or decrease the rates they pay providers across the board. The choices were met with a collective shrug at the January meeting. “Everybody on both sides [was] just like, ‘Yup. That’s what it is,’” she said.

Health economists have warned that higher health care costs may be an “unintended consequence” of AI scribes, which were initially deployed to help reduce physician burnout by documenting patient visits. But what if AI scribes raising patient bills means that they are working as intended? After all, the tools aren’t free, said Tinglong Dai, a professor at Johns Hopkins University’s business and nursing schools, and hospitals need to recoup the cost of deploying and maintaining the scribe technology.

Health system executives, originally thrilled by how much happier their physicians are because of the AI scribes, are now looking for financial value out of the tools. Mac Boyter, a senior research director at health care research firm KLAS, told STAT that one chief medical information officer described AI scribes in 2023 as “magic” that brought the joy back to medicine. But now, she is telling him that while that’s great, she needs the technology to provide “coding uplift” — higher reimbursement for patient visits — along with other functions like ordering follow-up lab tests or putting in prescriptions.

“Every CIO I talk to, every CMIO of health systems, every CFO, is saying, ‘Yeah, our clinicians love ambient tools. But I need to see the money. We need to see revenue generation or cost savings, we need to see operational efficiencies, workflows,’” said Boyter.

AI is going to increase health care costs rapidly until the industry or policymakers take action, said Pearson. “Right now, ambient scribes are inflationary, and that’s a problem. We need technology to help us lower health care costs.”

How AI scribes drive up medical bills

There are three reasons why AI scribes increase the amount doctors are billing, said Bobby DuPre, chief medical information officer of Louisiana-based FMOL Health’s ambulatory groups.

One is that burnt-out providers often only document the bare minimum necessary to send off a bill, and that usually ends up at a moderately complex, or Level 3, visit. But once they got access to the AI scribe, the comprehensive, automatic documentation showed that the providers had actually been doing visits that were complex or long enough to be billed at Level 4 or Level 5, he said. At FMOL Health, using Suki’s AI documentation tool led to a 6.5% increase in Level 4 visits, translating to an extra $862 per clinician per month.

Another reason why health systems are seeing a bump is because scribe software like Suki’s often also suggests codes. “Now you have a tool that’s saying, ‘Hey, you talked about their UTI, but you didn’t add it to your visit diagnosis. Do you want to add it now? And if you add it, I’ll give you an assessment and plan based on what you talked about with the patient,’” DuPre told STAT.

A third reason is volume. When physicians aren’t manually taking notes from patient visits and updating their electronic health records, they end up with more time. Even though health systems generally aren’t pressuring doctors to use their freed-up time to see more patients, it seems doctors might be doing that anyway.

At FMOL, a KLAS study found that after starting to use Suki, clinicians saw 22% more patients. Most ambulatory clinicians are on a reimbursement model that “means the more people they see, the more payment they get,” said DuPre. That incentivizes that outcome, but also means more patients can get care. In a study at the University of California San Francisco, physicians using an AI scribe brought in approximately $3,000 extra dollars per year.

Blue Cross Blue Shield of Massachusetts told STAT that its costs per visit have risen 30% since 2021, likely due to providers’ use of AI tools. It’s unclear how much of that is actually due to AI. Still, the problem has gotten so bad that many health insurers have instituted programs that identify “outlier” bills for Level 4 and Level 5 visits and pay for a visit one level lower. On average, a spokesperson for Cigna told STAT, this is a difference of $50 per visit.

Suki, who says that its tools can increase health care provider revenue by $1,688 per user per month, declined to comment on insurers’ assertion that its AI tools are partially responsible for driving up the cost of care.

How does it affect the health care system?

Health economists warned that this AI coding arms race — fueled by AI scribes and autonomous coding tools maximizing codes on one side, and by insurer algorithms trying to minimize payments on the other — is a zero-sum game that could really hurt some of the most vulnerable providers.

Perhaps providers’ bills for care are just becoming more “accurate” due to the AI scribes more faithfully recording the patient visit. But if the underlying care isn’t changing, insurers have a fiduciary duty to their beneficiaries to lower their rates, said Daniel Polsky, a health economics professor at Johns Hopkins.

“Vendors are selling the software to providers as if it’s going to make them rich and they want to get a cut of it,” he said. But if the payers eventually lower their rates so that providers end up getting about the same amount anyway, “then the providers paid the vendors too much for something that wasn’t really adding real value.”

DuPre disagrees; he thinks the scribes are boosting the quality of care provided. He’s still a practicing rheumatologist and described a patient he’s been seeing for the last 15 years, since he finished his fellowship. Around the third time he used the AI scribe in a visit with the patient, she told him, “This has been the most engaged you have ever been since we’ve been seeing each other,” he said.

The value isn’t just in paying more attention to the patient, either — he now summarizes the whole visit out loud, for the benefit of both the AI and the patient, which has helped patients bring up issues they didn’t cover and makes them feel more taken care of.

He’s not worried about AI scribes ballooning health care costs. “When you look at the breakdown of health care costs, physician payment is not the large balance of that cost. It really is not. It comes down to hospital, acute care, testing, new tests,” he said. He argued that AI scribes might even lower costs in the long run because if providers are seeing 22% more patients, they might keep people out of the emergency room or hospital.

Health economists were skeptical of this argument. While Polsky agreed that a single tool isn’t going to raise costs exponentially, he pushed back on the idea that this is too small of an issue to worry about. “Getting bang for our buck in health care will require attention to every part of health care not delivering value,” he said in an email. “This is the attitude that has created a low value health system. We should all be responsible stewards in our part of the health system elephant.”

Dai was more worried about the downstream effects of the coding war. What if, as in the recent study from Blue Cross Blue Shield, insurers demand to see treatment to affirm that a diagnosis affixed to a bill is legitimate, and that prompts clinicians to provide inappropriate care to justify the higher bill? What if a primary care physician who might not otherwise have referred a person to a specialist refers them, triggering more unnecessary testing and procedures and preventing people who need that specialist from being able to book an appointment?

That, Dai said, “can create a cycle of overtreatment on one side and tighter restrictions on the other, making the system more costly and harder to navigate for everyone.”

What will happen?

Polsky said that if we need to pay doctors more, we should just figure out how to pay doctors more, and uncouple it from the use of AI. Polsky, Dai, and Pearson all see insurers reducing reimbursement rates across the board as an inevitable outcome. What happens when that comes?

Health systems, typically those caring for underserved populations, that haven’t had the time or resources to implement AI will get even less money, said Polsky. “The winners will be the most sophisticated providers that can use these tools most effectively. They’re not gonna be the downtrodden single PCP that is taking care of a neighborhood that is [designated as] critical access care,” he said. “That’s not where these technologies are being adopted. They’re being adopted [by] the haves, not the have-nots.”

Pearson said that AI is being layered on top of “existing adversarial workflows,” making it cheaper for providers to maximize bills and for insurers to tamp down on spending, making the overall situation worse. While we could devise a better and more efficient system, “no one in the existing system has a lot of incentive to do that,” she said.

One such alternative could be value-based care. But that, in turn, has its own problems.

Access this article at its original source.