Peterson institute says healthcare AI spreads fastest where fee-for-service pays for it
Peterson Health Technology Institute says AI is spreading fastest in U.S. fee-for-service billing and claims tools. Without coordinated action from government and industry, its report warns, AI could make the system costlier and more wasteful, Axios reported.
The Engineer · Build desk
Pearson's roles: policymakers, regulatory and coverage rules; tech companies and investors, evidence, monitoring and safeguards; payers, payment models that encourage adoption without increasing costs. Providers may lack incentive to invest in clinical AI with no viable payment model.
- decision Policymakers Should address regulatory requirements and coverage rules, Pearson said, claim 16
- decision Tech companies and investors Should focus on evidence, ongoing monitoring and safeguards, Pearson said, claim 16
- decision Public and private payers Should design payment models that encourage adoption without increasing costs, Pearson said, claim 16
- exposure Providers May lack the incentive to invest in and scale clinical AI that has no viable payment model, claim 15
| Who | How | Kind | Claim |
|---|---|---|---|
| Policymakers | Should address regulatory requirements and coverage rules, Pearson said | decision | 16 |
| Tech companies and investors | Should focus on evidence, ongoing monitoring and safeguards, Pearson said | decision | 16 |
| Public and private payers | Should design payment models that encourage adoption without increasing costs, Pearson said | decision | 16 |
| Providers | May lack the incentive to invest in and scale clinical AI that has no viable payment model | exposure | 15 |
What happened
- The billing and claims applications spreading fastest have drawn attention for fueling bot wars and rising healthcare costs.
- Outcome-based and capitation models shift incentives toward results, but the report says they underpay for workflow redesign and other upfront investment.
- PHTI concludes that no single payment model fits every AI application and clinical setting.
Why it matters
- cost Payers who answer the reimbursement gap by paying for each use of AI risk a sharp rise in health spending, and payers and patients carry that bill.
- exposure Administrative AI tools are unregulated and profitable for the organizations deploying them, so little outside pressure slows their spread.
- constraint Clinical applications with no viable payment model may never get the provider investment needed to scale, whatever their clinical promise.
- decision Pearson splits the work three ways: regulatory and coverage rules for policymakers, evidence and monitoring for investors and vendors, and new payment designs for payers.
The adoption pattern follows the payment rule. PHTI says the applications moving fastest are the ones compatible with fee-for-service, including administrative tools for billing and processing insurance claims [3]. The report says current financial incentives already reward more services and more administrative complexity [2]. In our view, a tool that fits an existing incentive has the shortest route to a purchase order.
Caroline Pearson, PHTI's executive director, said these administrative tools are unregulated and can bring profit to the organizations that deploy them [11]. She added that the current system rewards less efficient human labor, because that is the only reimbursement model [12]. A tool that pays for itself and answers to no regulator sells easily.
Clinical tools meet a different set of obstacles. The report says clinical AI could separate the volume of services from staff headcount and capability, scale services at lower marginal cost, and let clinicians monitor more patients [5]. Fee-for-service rewards providers for patient visits, while AI can separate a clinical result from clinician time and support continuous long-term care [8]. PHTI says common payment models are poorly suited to paying for services that AI delivers or assists [9].
The obvious fix has its own hazard. Paying per use of AI could sharply raise health spending [10]. Outcome-based and capitation models shift incentives from volume to results, but the report says they underpay for workflow redesign and other upfront investment [13]. PHTI concludes that no single payment model suits every AI application and clinical setting [14].
The evidence side is also thin. The report says standards for evaluating evidence, safety, outcomes, effectiveness and post-deployment monitoring are underdeveloped, so regulators, buyers and users have no clear way to tell what works [6]. Existing healthcare rules were not written for AI, which leaves liability, accountability, oversight and the limits of professional practice open. Clinical and administrative workflows were not designed for autonomous or AI-involved processes either [7].
Pearson said that if the sector does not act differently, it will lay technology over workflows that are already broken, costs will rise, and the expected benefits will not arrive [17].
Axios obtained the report first [18]. The account we have is mezha.net's relay of that story, and it includes no adoption figures or spending estimates. "Fastest" is therefore PHTI's characterization, not a measured share of deployments.
What to watch
- Whether public or private payers publish reimbursement designs for AI-delivered services that avoid pay-per-use, the condition Pearson set for payers.
- Whether policymakers act on the regulatory requirements and coverage rules Pearson named, since administrative AI tools are currently unregulated.
- Whether the full PHTI report puts adoption or spending figures behind the claim that billing and claims tools lead.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence35
- Adoption
- Insufficient
- Hype gap
- Insufficient
- Incentives40
- Confidence50
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
A new Peterson Health Technology Institute report warns that without coordinated action by government and industry leaders, AI could make the U.S. healthcare system costlier and more wasteful.
- [2]
The report says current financial incentives in healthcare encourage higher service volume and more complex administrative processes.
- [3]
The fastest-spreading AI applications are those compatible with fee-for-service payment, including administrative tools for billing and processing insurance claims.
- [4]
These applications have drawn attention for fueling bot wars and rising healthcare costs.
- [5]
AI can separate the volume of services from staff headcount and capability, scale services at lower marginal cost, and help clinicians monitor more patients, respond faster and spend more time on the human side of care.
- [6]
Clinical AI adoption is held back by underdeveloped standards for evaluating evidence, safety, outcomes, effectiveness and post-deployment monitoring, so regulators, buyers and users have no clear way to determine what works.
- [7]
Existing healthcare rules were not created for AI, leaving open questions of liability, accountability, oversight and limits of professional practice; clinical and administrative workflows were also not designed for autonomous or AI-involved processes.
- [8]
Fee-for-service rewards providers for patient visits, whereas AI can separate a clinical outcome from clinician time and labor and enable continuous long-term care.
- [9]
According to PHTI, common payment models are poorly suited to paying for services provided by or with the help of AI.
- [10]
Paying for each use of AI could sharply increase healthcare costs.
- [11]
Caroline Pearson, PHTI executive director, said administrative AI tools are unregulated and can bring profit to the organizations that deploy them.
- [12]
Pearson said the current system therefore incentivizes less efficient human labor, because it is the only reimbursement model.
- [13]
Outcome-based and capitation payment models shift incentives from service volume to results, but the report says they do not adequately reward workflow redesign and other upfront investment in clinical AI.
- [14]
PHTI concludes that no single payment model fits all AI applications and clinical settings.
- [15]
If promising clinical applications lack a viable payment model, providers may lack the financial incentive to invest in and scale them.
- [16]
Pearson said policymakers should address regulatory requirements and coverage rules, technology companies and investors should focus on evidence, ongoing monitoring and safeguards, and public and private payers should design payment models that encourage adoption without increasing costs.
- [17]
Pearson said that if the sector does not start acting differently, it will simply lay technology over already broken workflows, everything will become costlier, and the expected benefits will not be realized.
- [18]
The PHTI report was first obtained by Axios.
Get this ledger inside ChatGPT. Ask about any of our stories and the answer brings its figure, each claim’s verdict and the sources. Free, no account.
Sources
1 independent publisher whose own reporting we read for this story.
- mezha.netПерешкоди на шляху до реалізації потенціалу ШІ в охороні здоров’я
1 article · October 14, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.
Topics
- Healthcare payment modelsFollow
- Healthcare AIFollow
- AI RegulationFollow