Episode 3 of The 80 Million Podcast: AI and Digital Innovation in Medicaid — Promise, Peril and What to Buy
Medicaid agencies face critical decisions about what technologies improve care, expand access, and lower costs. A conversation with Peterson Health Technology Institute helps cut through the noise.
Author: Jared Augenstein, Patti Boozang and Caroline Pearson
Editor: Amanda Eisenberg
tl;dr
In Episode 3 of The 80 Million Podcast, host and 80 Million editor Patti Boozang speaks with Caroline Pearson, executive director of the Peterson Health Technology Institute (PHTI), and Jared Augenstein, senior managing director at Manatt Health, about what Medicaid leaders should believe — and question — about the explosion of artificial intelligence (AI) and digital health solutions entering the market.
AI and digital health tools hold real promise for expanding access and improving care in Medicaid, particularly in mental health, opioid use disorder and chronic disease management. But technology is advancing far faster than the evidence base needed to guide public investment decisions.
AI is not yet delivering administrative savings. In today’s fee-for-service system, AI is intensifying coding optimization, prior authorization disputes and administrative volume rather than reducing overall costs. That means that payment reform, not technology alone, is the critical policy lever if states want AI to improve affordability.
The $50 billion Rural Health Transformation Program (RHTP) is a rare opportunity for states to make strategic, evidence-based technology investments and ensure safety-net and rural providers are not left behind in the next phase of digital transformation.
The 80 Million Impact
State Medicaid programs are under extraordinary pressure as they navigate federal funding uncertainty, H.R. 1 implementation, and health care cost growth that consistently outpaces both inflation and state revenue growth. States have myriad technology vendors pitching solutions to help alleviate those burdens, particularly around helping drive down the costs associated with certain clinical conditions and administrative functions. These technology solutions, increasingly, are AI powered and promise to be the differentiator for patients and Medicaid budgets alike.
It can be daunting for state Medicaid leaders to evaluate which technologies deliver, where there are risks, and the types of structural changes that are needed for innovation to benefit the people Medicaid serves rather than the vendors selling to it.
Digital Health Tools: Identifying What Works
The integration of new digital technologies into health care promises to drive three key goals: improve health outcomes, lower costs or expand access — goals that are critical to the Medicaid program.
Our discussion with PHTI centered on a critical point for Medicaid leaders: the evidence base for digital health is finally becoming mature enough to separate technologies that meaningfully improve care from those that primarily generate hype. PHTI’s independent evaluations of digital health tools show that some categories are delivering measurable clinical value and expanded access — while others have failed to demonstrate meaningful improvement despite years of investment and market enthusiasm.
Digital mental health is one of the strongest examples of where the evidence is increasingly compelling. PHTI and other independent reviews have found that self-directed digital mental health tools — including AI-enabled cognitive behavioral therapy and conversational tools designed to support therapy delivery — can effectively improve outcomes for people with mild to moderate depression and anxiety. For Medicaid programs facing severe behavioral health workforce shortages, these tools offer something particularly important: scalable access. They can extend support beyond traditional clinic hours, reduce wait times and reach individuals who might otherwise receive no treatment at all.
PHTI has also identified strong evidence supporting digital opioid use disorder treatment models, particularly platforms that enable virtual medication-assisted treatment and ongoing patient engagement. These models can reduce transportation and geographic barriers that often prevent Medicaid beneficiaries from accessing or maintaining treatment. Similarly, virtual physical therapy has demonstrated clinical outcomes comparable to in-person care for many patients, while improving convenience and adherence.
But the evidence also underscores an equally important caution: not every digital health category delivers on its promises. Digital diabetes management is one of the clearest examples. PHTI’s evaluation found that digital diabetes solutions produced limited clinical benefits across broad populations, though meaningful improvements were observed in specific subgroups, particularly patients newly starting insulin. For most patients, digital solutions did not outperform routine clinical management, behavioral interventions and lifestyle modifications.
The lesson for Medicaid programs is clear: enthusiasm for innovation cannot substitute rigorous evidence evaluation. States will need disciplined procurement, ongoing outcomes measurement, and clear expectations for value if they want digital health investments to improve care rather than simply add another layer of cost and complexity.
Importantly, states may soon have a once-in-a-generation opportunity to shape how this transformation unfolds. The new $50 billion in RHTP funding could provide states with meaningful flexibility and capital to make strategic, evidence-based investments in digital infrastructure, AI-enabled care models, and technology adoption across Medicaid delivery systems.
AI and Administrative Efficiency
The optimistic narrative about AI and administrative efficiency in health care, that administrative savings are “low-hanging fruit,” is not materializing in practice. In today’s market, AI is increasingly being monetized in an arms race between payers and providers, and the near-term effect is not administrative simplification, but administrative escalation — more transactions, more friction and higher system costs.
One of the clearest examples is ambient scribe technology, now widely deployed to generate more complete visit documentation. Better documentation can improve clinical records — but it also predictably increases coding intensity by supporting more billable diagnoses and higher-complexity visit levels. That may improve provider reimbursement accuracy, but it also increases spending for payers and patients unless counterbalanced by payment design and oversight.
States should also closely track AI’s impact on prior authorization and utilization management. AI is making it cheaper and faster to generate denials, generate appeals, and generate counter-appeals — turning what used to be a single back-and-forth into multiple rapid cycles, often without a deliberate decision by a human to escalate. The result is a classic “efficiency for each actor, higher cost for the system” dynamic: Each side can automate its administrative posture, but the net effect is increased administrative intensity and cost.
The underlying structural issue is the payment model itself. In a fee-for-service environment, broad adoption of AI for administrative processes and revenue cycle functions is more likely to increase total spending than reduce it because it strengthens each actor’s ability to optimize within existing incentives. That means that states can’t treat AI adoption as merely a technology strategy; it must be paired with payment reform. The promising path is for states to move providers and plans into value-based, global budget, and capitated payment arrangements where the economic incentive shifts from maximizing units of service to improving outcomes and managing total cost of care. In these models, AI and other digital tools become levers for prevention, care coordination, risk stratification, workflow efficiency and reduction of unnecessary utilization — not simply revenue maximization.
The Bottom Line
Digital health tools and AI may transform Medicaid, but technology alone will not solve the program’s affordability, access or administrative challenges. The impact of AI will depend following the evidence on what works — and on reforming the payment incentives embedded in the health care system itself.
Listen to the full conversation on Spotify, Apple Podcasts or wherever you get your podcasts for a deeper discussion on how states can cut through the AI hype cycle and make smarter, evidence-based technology investments for Medicaid.
And don’t forget to subscribe to The 80 Million Podcast.