There are many ways to say “no” in medicine.
“No, this antibiotic won’t help a viral infection.”
“No, this test won’t change our management.”
“No, that surgery carries more risk than benefit.”

And then there is the newest way:
“No, the algorithm doesn’t prioritize your health.”

Welcome to WISeR, short for “Wasteful and Inappropriate Service Reduction,” a Medicare pilot program that promises to usher artificial intelligence into health care decision-making with all the subtlety of a parking ticket and all the moral imagination of a spreadsheet. According to its architects, WISeR will save taxpayers money, protect beneficiaries, reduce fraud, streamline care, and, almost as an afterthought, continue to provide appropriate medical services.

It is already being called an “AI death panel.” CMS objects to that label. Naturally. One does not name one’s own guillotine.

Target practice: Aim for the elderly, fire at the knees

Every great experiment needs a soft launch. WISeR wisely begins with “low-hanging fruit,” procedures that are common, expensive, and disproportionately affect older adults. Knee arthroscopy for osteoarthritis. Electrical nerve stimulators. Skin and tissue substitutes. The elderly, in other words, have been selected for beta testing.

[Image of knee arthroscopy for osteoarthritis]

Critics worry that this is a slippery slope toward life-and-death decisions being mediated by AI. CMS reassures us that WISeR does not involve “life-threatening” care. True enough. Knee pain won’t kill you. It will simply ensure that the last decade of your life is conducted in slow motion, punctuated by falls, addicted to opioids, and full of regret.

If the real goal is cost containment, one wonders why CMS stopped here. Why not be honest? Why not expand euthanasia and really bend the cost curve? Canada has already demonstrated the power of policy branding through its national program, Medical Assistance in Dying (MAID). Think of the savings. Think of the efficiencies. Think of the acronyms (see below).

“666”: Six states, six vendors, six years

WISeR will be piloted for six years in six states (Arizona, New Jersey, Ohio, Oklahoma, Texas, and Washington), each assigned a different private contractor wielding its own proprietary AI tools. This will ensure six different interpretations of “medically necessary,” six flavors of denial logic, six proprietary black boxes, and six unique ways to tell clinicians they didn’t fill out the form correctly.

CMS is trying to convince us that this pilot will ensure innovation, competition, and learning. To me, “666” feels less like innovation and more like a pact with the devil. Consistency is overrated anyway. Medicine thrives on variability, especially when it’s algorithmic.

Why not skip the pilot altogether? Roll it out nationally. Let every American experience the thrill of wondering whether their care depends on geography, vendor selection, or whether the AI had a good training set that morning.

Nothing builds public trust like randomness.

For-profit denial: Now with incentives

WISeR’s most elegant innovation is its payment structure. Contractors are paid a percentage of the money they save Medicare by denying care. CMS calls this “averted expenditures.” Normal people call it a bounty system.

This is not a bug. It is the feature.

The contractors only get paid if care is not delivered, or if a denial survives appeals long enough to ripen into profit. CMS assures us that licensed clinicians will review denials, which is comforting in the same way it’s comforting to know that someone technically glanced at your tax return before auditing you.

Ethics, curiously, do not appear as a performance metric.

Branding matters: WISeR, but not smarter

“Wasteful and Inappropriate Service Reduction” is a mouthful, but WISeR is a strong acronym, suggesting prudence, restraint, and sagacity. It pairs nicely with CMS press releases about creativity and stewardship.

But acronyms cut both ways. If WISeR exists, it deserves a competitor:

• SLoP: “Statistical Logic over Patients”
• DUMB: “Data-Uninformed Medical Bureaucracy”
• NOPE: “Necessary Outcomes Preemptively Eliminated”
• CARE-LESS: “Computational Authorization for Reducing Expenses, Leaving Seniors Sore”

Simply call it STUPID: “Systemic Triage Using Predictive Inference for Denials.” Because branding is important when you’re automating denial.

On second thought, that acronym won’t work. You want the name to sound thoughtful, not predatory.

“Strategic Tech-Enabled Uncareful Pruning of Indispensable Decisions” didn’t test well in focus groups.

Transparency theater: Trust us, it’s proprietary

CMS insists that WISeR will be transparent. Providers, however, will not see the algorithms, the weighting of variables, or the decision logic. That information is proprietary. You wouldn’t want clinicians understanding how medical decisions about their patients are made. That could lead to questions.

We are told that AI will merely “expedite” reviews and “support” clinicians. This is the same reassurance offered every time a human role is quietly hollowed out and replaced with a dropdown menu.

The experience of Medicare Advantage should have taught us something here. AI-driven prior authorization in that program has been associated with widespread inappropriate denials, delays in care, and class-action lawsuits. CMS seems to argue that the problem was never the tools, only who was holding them, and the government under traditional Medicare is WISeR than private insurance companies administering Medicare Advantage plans.

The roundtable of trust (Spelled T-R-U-S-T)

Perhaps the most charming detail in the WISeR saga (nothing is carved in stone yet) is that CMS held a roundtable with insurers shortly before introducing the program (on June 27, 2025), after which it proudly announced that insurers had pledged to fix the “broken prior authorization system” in Medicare Advantage.

A pledge. From insurers.

This is the regulatory equivalent of asking the fox to pinky-swear not to eat the chickens, and then giving it a better lockpick.

CMS asks providers and patients to trust that lessons have been learned, incentives have been aligned, and guardrails will hold. Trust, after all, is cheaper than oversight.

Administrative burden: Because doctors needed more paperwork

WISeR promises to reduce administrative burden. This is technically true if you define “reduce” as “redistribute entirely onto clinicians.”

Providers must now choose between prior authorization upfront or pre-payment review later, both involving documentation, delays, resubmissions, peer-to-peer reviews, and appeals. Nothing says efficiency like doing the work twice, slowly, while your patient waits.

The algorithm does not experience burnout. The physician explaining to an 82-year-old why her knee surgery was denied by a vendor in another state does.

Conclusion: Automation speeds rejection, not healing

WISeR is not stupid because it uses AI. It is stupid because it applies AI to the wrong problem, with the wrong incentives, under the wrong assumptions, and then calls the result wisdom.

If the goal were truly better care, CMS would invest in primary care, physical therapy access, shared decision-making, and time. Those are expensive. Denial is cheaper.

So, we get WISeR: An experiment that mistakes efficiency for ethics, savings for stewardship, and automation for judgment. It will be fast. It will be scalable. And it will be very good at saying “no.”

Whether it is good for patients is, apparently, a secondary outcome.

Arthur Lazarus is a former Doximity Fellow, a member of the editorial board of the American Association for Physician Leadership, and an adjunct professor of psychiatry at the Lewis Katz School of Medicine at Temple University in Philadelphia. He is the author of several books on narrative medicine and the fictional series Real Medicine, Unreal Stories. His latest book, a novel, is Against the Tide: A Doctor’s Battle for an Undocumented Patient.