Primary Care Is Not a Substitutable Input

Hawaiʻi's largest insurers, and now an HMSA partner, are treating primary care as a substitutable input that AI can replace. The HHTF lays out the record and says the relationship is not for sale

HHTF Staff · 2026-06-14

The appointment is where the diagnosis is first heard, the medication list is reconciled, and the second question finally gets asked. Illustration: Hawaiʻi Healthcare Task Force.

What Hawaiʻi's insurers are saying about the doctor-patient relationship, and what the Task Force is saying back.

By David Isei, Executive Director, Hawaiʻi Healthcare Task Force

There is a counting story and there is a quieter story, and the people of Hawaiʻi deserve to hear both.

The counting story is the one already on the record. The 286 percent inflation. The 4,867 names. The 1,260 federally validated primary care physicians who actually exist in the state's largest commercial Medicaid network. That story is documented, and it stands on its own.

4,867 PCPs filed by HMSA in Med-QUEST Report 403 1,260 Validated after NPI, PECOS, APCD, and license reconciliation 286% Inflation versus the validated count

The quieter story is what the country's largest health insurers, and at least one major Hawaiʻi insurer that has now joined them, believe about primary care itself. It is not a story about counting. It is a story about whether the relationship between a patient and a primary care physician is something a company can engineer its way out of.

The argument moving through the conference rooms

For most of the past decade, a particular argument has been circulating among national health insurers. It goes like this. Primary care is expensive. Primary care physicians are scarce. The next generation of artificial intelligence can perform a growing share of the cognitive and administrative work that primary care physicians do today. Therefore, in the long run, the patient-physician relationship is a substitutable input.

That argument has not stayed inside the conference rooms. It has been made out loud, by named executives, on the record, in front of Wall Street.

On January 27, 2026, UnitedHealth Group's earnings call carried the argument national. Tim Noel, the chief executive of UnitedHealthcare, told investors the company anticipates "operating cost reductions of nearly $1 billion in 2026 — many AI-enabled" and that "over 80% of calls from members leverage AI tools" ( UnitedHealth Group Q4 2025 Earnings Remarks ). On the same call, Patrick Conway, MD, chief executive of Optum and a former director of the Center for Medicare and Medicaid Innovation, named the strategy directly: "AI-first new product innovation, strengthening Optum's care provider market offerings" ( UnitedHealth Group Q4 2025 Earnings Remarks ). Optum is the largest employer of physicians in the United States. Its chief executive told shareholders that the future of those physicians is AI-first.

Two months later, Optum Insight's chief executive Sandeep Dadlani put a finer point on the timeline. "Since the advent of generative AI, we've really doubled down on training, on investments, on driving meaningful use cases," he told STAT News in April 2026, in a piece that documented the use of AI to automate "everything from fraud detection, to clinical documentation, to the selection of billing codes that determine how much a given medical encounter costs — and who pays" ( STAT News, April 6, 2026 ).

This is not one company. The largest national payers are following the same playbook. Elevance Health's chief executive Gail Boudreaux told the House Ways and Means Committee in January 2026 that her company is "expanding electronic prior authorization to reduce paperwork and speed decisions, using technology, including AI" ( Congressional Testimony, January 22, 2026 ). The National Association of Insurance Commissioners reported in May 2025 that 84 percent of health insurers already use AI or machine learning in some capacity, and that between 68 and 82 percent already use AI to review prior authorization requests for approval ( NAIC Health Insurance AI/ML Survey Report, May 9, 2025 ).

The academic literature has been blunt about where this trajectory leads. Arch G. Mainous III, chair of family and preventive medicine at the University of South Carolina, wrote in Frontiers in Medicine in 2022: "If the decisions for modifying the healthcare system around new technology is driven by the desire of insurers and health systems to decrease costs, particularly labor costs, then the resulting system will likely need fewer and fewer primary care physicians" ( Frontiers in Medicine, April 26, 2022 ). Mainous documented that virtual-first primary care plans are now "priced up to 20% lower than other traditional plans offered by the same insurer," and that the savings come from reducing the need for physicians at all.

And then it arrived in Hawaiʻi

On September 18, 2025, this conversation reached our state.

On that date, HMSA announced it had joined Stellarus, an artificial intelligence platform housed under Blue Shield of California's parent company. The Stellarus platform, in its own promotional materials, advertises "fully automated, near real-time completion of transactions such as prior authorization," "member and physician access to a comprehensive digital health record that facilitates truly personalized care," and "technology-enabled, personalized support for patients with chronic conditions" ( PR Newswire, September 18, 2025 ). HMSA's chief executive Mark Mugiishi, MD, said in the release that the partnership would allow HMSA to "enhance our ability to innovate and grow our technological capacities by investing in Stellarus" so that "together, we can achieve remarkable advances and provide the best possible care for our community."

We do not doubt the sincerity of that aspiration. We are raising the operational question. A platform whose marketed feature is "fully automated" prior authorization is, by design, a platform that removes a layer of human review from the determination of whether a Hawaiʻi resident gets a procedure, a referral, or a medication. The Task Force is asking what that resident is supposed to do when the answer comes back wrong, when the algorithm misreads a chronic condition, and when the appeal pathway is itself routed through the same system that produced the denial.

We are also asking a more direct question. Has any executive at any Hawaiʻi insurance carrier, in any private or semi-public forum, argued that AI will replace some portion of primary care provider work in the next five to ten years? We will not name names here. We will simply say that the conversation is happening, in this state, in 2026, in the same offices that file the network adequacy reports we have just validated. Anyone who has been in those rooms knows what we are referring to.

What the Task Force does not accept

Primary care is not a substitutable input. It is the relationship through which the rest of the healthcare system is made to work. It is the appointment where a new diagnosis is heard for the first time, where a medication list is reconciled, where the patient who is afraid to ask a second question is given the moment to ask it. It is the kind of work that does not survive being optimized into a chatbot transcript and a denial code.

Talk to any family physician in Hawaiʻi and the point becomes concrete. Continuity is the doctor who remembers that the grandmother on three medications also cares for two grandchildren, and adjusts the plan to the life and not only the chart. Empathy is the pause that lets a frightened patient finish the sentence the algorithm would have truncated. That is not overhead to be automated away. In a state of islands, rural communities, and multigenerational households, it is the load-bearing wall of the entire system.

The professional consensus inside medicine is unambiguous. The American Medical Association's chief executive John Whyte, MD, MPH, stated in March 2026 that "it is critical that augmented intelligence be designed to enhance — not replace — physicians" ( AMA, March 12, 2026 ). The American College of Physicians, in a 2024 policy paper in Annals of Internal Medicine , put it plainly: "AI-enabled technologies should complement and not supplant the logic and decision making of physicians and other clinicians" ( ACP, June 4, 2024 ). The American Academy of Family Physicians, at its 2025 Starfield Summit, framed it as a choice: "We can either let AI happen to primary care, or we can shape how it happens with intention, equity, and purpose" ( Annals of Family Medicine, September-October 2025 ).

Patients are equally clear. Bain & Company's 2025 national survey found that only 28 percent of patients are comfortable with "AI becoming their doctor," and that AI chatbots rank lowest among care delivery preferences, with a net negative 46 percent score for sick visits and a net negative 53 percent score for chronic condition visits ( Bain & Company, October 22, 2025 ).

The distinction that matters

Our position is not nostalgic, and it is not anti-technology. The Task Force represents physicians, advanced practice clinicians, hospital leaders, rural health advocates, and community organizations, and we use machine learning ourselves to validate provider directories and surface network gaps that a list of paper names cannot reveal. The question was never whether AI belongs in healthcare. The question is what AI is being used to do, and to whom, and on whose authority.

When AI checks the accuracy of a payer's network filing, the people of Hawaiʻi are better off. When AI substitutes for the primary care relationship itself, they are abandoned, with a more sophisticated abandonment than an inflated directory produces.

The Hawaiʻi Healthcare Task Force strongly disagrees with any insurance plan, in Hawaiʻi or on the mainland, that proposes to replace human primary care relationships with artificial intelligence. We will say so in every public forum, in every legislative hearing, and in every conversation with state regulators, until the proposition stops being floated.

Hawaiʻi Health Intelligence Platform Clinically Integrated Network (HHIP CIN) is in pre-formation. Statements regarding founding members, payers, and partnerships describe targets and candidates, not signed agreements. Nothing in this document constitutes legal, tax, regulatory, clinical, or investment advice; an offer to enter into a contractual relationship; or representation of an existing partnership. All requirement statuses reflect the project team's good-faith assessment as of the date shown.