PHTI’s Report on Administrative AI

Peterson Health Technology Institute (PHTI) convened health care leaders to examine how AI is reshaping healthcare administrative processes.

WHAT: In January 2026, the Peterson Health Technology Institute (PHTI), a non-profit organization that evaluates digital health technologies, convened senior leaders from health systems, health plans, technology developers, investment firms, and federal agencies to discuss how AI is being deployed in healthcare administrative processes. PHTI’s takeaways were published in a report entitled: “Administrative AI: Current Use and Potential Impact.” The report focused on two administrative use cases where AI adoption is accelerating: (1) prior authorization and (2) medical billing. The workshop was held under the Chatham House Rule to encourage candid dialogue.

IN BRIEF: AI is being deployed aggressively on both sides of the prior authorization and medical billing transactions – but the results are more activity, not less cost. PHTI’s report highlights four realities shaping the current administrative AI landscape:

1. AI may reduce costs for individual organizations managing prior authorizations, but it has not reduced overall system-level costs – and risks creating “bot wars” that multiply back-and-forth communications without improving efficiency;
2. Provider adoption of AI scribing and coding tools is increasing billing intensity and inflating medical spending, with payers citing “aggressive” coding as a driver of higher medical expenditures in 2025 and early 2026;
3. Health plans are beginning to respond to AI-driven billing increases with across-the-board downcoding and other reimbursement reductions – but the full impact of these policies is uneven and not yet understood; and
4. Reimbursement policy reform is the strongest lever to drive administrative efficiencies and system-level cost savings – technology alone cannot fix processes built on misaligned incentives.

WHEN: PHTI published the report in April 2026. It is the second installment in a three-part workshop series; future sessions will address payment models for AI. This report follows PHTI’s first workshop on clinical AI evidence and policy requirements (February 2026). See Maverick’s report here.

PHTI focused on prior authorization and medical billing because these are the two administrative processes where AI is being deployed most aggressively. There are several takeaways from PHTI’s report:

Prior Authorization

Theme 1: AI may reduce individual organization costs, but not overall system-level costs.

• While these tools help each side move faster, participants raised concern that optimizing each side independently risks making the overall process more activity-intensive rather than more efficient. As one participant put it: “Bots don’t get tired of asking questions, so my review queue keeps growing.”
• There is no evidence yet that AI adoption translates to lower average cost per claim once the cost of the AI solution itself is factored in.

Theme 2: Real-time prior authorization at the point of care is emerging but not yet scalable.

• Emerging solutions—such as the Optum Real platform and a Highmark-Abridge collaboration—are designed to complete the full authorization process during the patient visit by integrating clinical documentation, policy matching, and authorization determination in real time.
• Early pilots are promising: one reported an 88% reduction in appeals and a 68% reduction in denials caused by missing information. But the gap between narrow pilots and a production-ready, generalizable solution is substantial.
• Scaling real-time adjudication across a broad, multi-stakeholder market will likely require further policy intervention to reconcile fragmented medical policies and build infrastructure for real-time, bidirectional data exchange.

Theme 3: Data standards and digitization can reduce information asymmetry, but AI’s impact is limited by variation across medical policies.

• The CMS Interoperability and Prior Authorization Rule (CMS-0057-F), finalized in 2024, requires implementation of prior authorization APIs by January 1, 2027 – enabling providers to determine requirements and receive decisions electronically in real time. See Maverick’s brief here
• In June 2025, more than 50 health plans committed to voluntary reforms including reducing the volume of services subject to prior authorization and expanding real-time responses to at least 80% of electronic requests by 2027.
• PHTI participants noted that digitizing the exchange of medical policy information does not standardize the content of those policies. Each health plan defines medical necessity according to its own clinical guidelines and benefit design. See Maverick’s brief on ePA proposed rule here.

Theme 4: AI is exposing deeper structural limitations in the prior authorization process that technology alone cannot fix.

• Participants raised a foundational question: before investing in technology to make a flawed process run faster, should the process itself be redesigned? AI is surfacing — and in some cases amplifying — inefficiencies that were always present but easier to overlook at lower transaction volumes.
• Structural alternatives discussed include prepayment review (assessing appropriateness after care delivery rather than before) and offering providers a discounted payment rate to bypass upfront review entirely. Both would represent a fundamental restructuring of health insurance business models and would require significant policy and industry alignment to implement.

Medical Billing

Takeaway 1: Provider deployment of AI is increasing billing intensity and inflating medical spending.

• AI scribes and automated coding tools are now standard across health systems. In a recent survey of large health systems, all reported adopting AI tools for clinical documentation.
• These tools enable more complete capture of clinical complexity, resulting in higher-complexity billing codes. One multihospital health system found that after deploying an AI scribe, documentation improvements led to a 5% increase in Level 5 encounters and a 7% increase in Level 4 encounters for established patients—translating to an average revenue increase of $1,004 per provider per month.
• Payers are citing “aggressive” provider coding as a driver of higher medical expenditures in 2025 and early 2026.

Takeaway 2: Health plans are responding with across-the-board downcoding, but the full impact is not yet known.

• Reported health plan responses include deploying algorithms to automatically adjust outlier high-complexity evaluation and management (E/M) codes, reducing reimbursement for select modifiers, and benchmarking submitted E/M levels against peer providers with similar patient populations.
• These responses are generating pushback from providers and state legislators. Several health plans have paused or rescinded policies following provider opposition; Missouri and Indiana introduced bills in the 2026 legislative session to restrict AI-enabled downcoding by health plans.
• Health plan responses may disproportionately harm providers that have not adopted AI documentation tools—who are more likely to be smaller, rural, critical access, and independent. These providers may face revenue reductions despite having not benefited from AI-driven billing gains.

Takeaway 3: Reimbursement policy is the strongest lever to drive administrative efficiencies and system-level cost savings.

• Current health plan responses are not sufficient to address AI-driven medical inflation; a coordinated policy response is needed. Participants suggested near-term guardrails and longer-term payment reforms, including:
  • Transparency requirements: require disclosure of when AI tools are used to generate or support medical coding;
  • Oversight and monitoring: implement cost growth targets and audits to detect inflationary coding trends; and
  • Price and coding adjustments: in the near term, adjust prices for practices exhibiting sudden increases in coding intensity; in the longer term, revalue CPT codes, RVUs, and DRG payment levels to reflect how AI is changing the effort and complexity involved.
• Without deliberate policy action, AI risks exacerbating current inflationary trends in medical billing. This makes revisiting how we pay for healthcare services all the more urgent.