A cautionary tale on deploying AI in healthcare without letting our mental muscles atrophy
As you know I’ve been paying attention to the invisible gravity of AI technology, the subtle force that baits us toward non-human support and efficiency before we’ve fully figured out what we’re losing as a trade-off. Lately, that conversation has hit a massive tipping point. Two completely different headlines landed on my desk and combined they magnified a friction point we've been talking about across the healthcare landscape.
First, Pope Leo XIV released a massive encyclical, Magnifica Humanitas, about keeping human agency at the front of AI. Not long after, digital health pioneer Dr. Eric Topol published a concern about what happens to doctors when we hand over too much control to algorithms.
Whether you’re looking at global ethics from the Vatican or trying to figure out how to deploy AI for prior authorizations and formulary exclusions along with other cognitive tasks, we're fighting the exact same internal battle: How much do we want to let AI do for us and how do we stop it from making us dumber? The good news is that there’s a sharpening awareness of these unexpected downsides. We aren't anti-AI; we're just realizing that we need to build a better navigation system to protect our own minds while we use it.
Before we lose it.
The throttle dilemma
To make sense of what’s happening to our teams, I keep coming back to the difference between a regular bicycle and an e-bike. An e-bike is an amazing piece of machinery. With pedal-assist e-bikes, many won’t engage the electric motor unless the rider is pedaling, supplementing your own leg power. It helps you go further, climb steeper hills, and tackle challenges that would normally exhaust you. Your muscles are still working, and your physical capacity grows, even if it is less than a traditional bike where the only thing powering you is you.
But if you twist that hand throttle to 100% and stop pedaling, everything changes. You aren't exercising anymore; you’re just a passenger sitting on a clunky bike frame bulging with an obvious battery. Over time, your leg muscles will atrophy. You might get to your destination faster, but you’ll be weaker when you get off.
Are we, in pharmacy and managed care operations, treating AI like an autonomous scooter instead of a pedal-assist e-bike? Are we letting the machine do the heavy lifting, completely forgetting that our human brain and our innate value as experienced executives and clinicians lies in the strength of our individual mental muscles? The fact that we are starting to notice this now is a great thing, as it indicates we have an opportunity to fix our expectations of AI before our collective judgment and knowledge weakens.
The three ways we lose our edge
In his analysis, Dr. Topol coins three terms to describe how we lose our skills when we rely too heavily on AI. Although he was talking about doctors, these traps apply just as heavily to managed care professionals, clinical pharmacists, and data analysts. Frankly, all of us.
Deskilling: This is what happens when you let a tool do a job for so long that you literally forget how to do it yourself. Topol points to gastroenterologists who rely on AI to find polyps during colonoscopies. When you take the AI away, the clinicians perform statistically worse than they did before the technology existed. In our world, we see this when veteran underwriters rely so blindly on automated financial models that they lose their intuitive feel for market anomalies.
Never-skilling: This is the scariest one for the next generation. It happens when junior analysts or pharmacy students use AI so early in their careers that they bypass the grueling fundamentals entirely. If an AI writes your data summaries and pipeline briefs from day one, you never get the core principles down. Professional intuition is built on the pattern recognition of past mistakes. If AI eliminates the mistakes, it also eliminates the wisdom and the often under-appreciated opportunity to learn from those mistakes.
Mis-skilling: This is simply using the tool for the wrong problem. We see this when leadership teams take an algorithm trained for basic administrative speed and expect it to handle nuanced, patient-facing care coordination or delicate clinical exceptions. It’s mistaking a data-cruncher for an empathetic human judge. This is mostly seen in patient-facing interactions.
My take? The threat isn’t that AI is going to maliciously replace us. The threat is that we are voluntarily handing over the exact analytical friction that makes us valuable in the first place. If we don't pedal, we don't learn.
The Pope’s reminder about discernment
This is where the Pope’s document, Magnifica Humanitas, hits home for corporate leaders. His core argument is that AI can imitate human intelligence, but it can’t experience relationships, responsibility, or ethical friction, to say the least. Human beings are dynamic, individual, and uniquely designed to express intelligence no AI can replicate.
“In the era of artificial intelligence, when human dignity is threatened by new forms of dehumanization, ours is the pressing duty to remain profoundly human.”
—Pope Leo XIV, Magnifica Humanitas, May 25, 2026
The Pope warns against a dangerous abdication of human agency, the moment we let an opaque algorithm make the final call because it's easier than thinking it through ourselves. For a pharmacy executive, this means remembering that an algorithmic approval or denial code is no substitute for a clinical conscience. When we are dealing with high-cost orphan drugs or complex oncology regimens, the final accountability has to stop with a human. Discernment is an exclusively human asset.
We are being hypnotized into laziness in the desire for efficiency, speed, and bottom line results.
How to take back our agency
I am optimistic about where we are heading, but we have to be intentional. We can easily navigate these unforeseen conditions by building human agency right back into our daily workflows.
Here are three simple rules you can start using with your teams to keep AI from eroding your competitive edge:
Write the first draft with your brain: Whether you're building a new underwriting framework or a clinical pathway, mandate that your team maps out the logic and hypotheses on their own first. Bring AI in during phase two to stress-test your assumptions, clean up the formatting, or pull data. Never let the machine dictate the blank page.
Create AI-free zones for training: When you're mentoring new hires or junior analysts, give them projects where AI assistance is completely turned off. Make them do the manual spreadsheet math and look up the data feeds natively. They need to build the muscle memory of the basics before you hand them the e-bike throttle.
Audit your tech stack quarterly: Take a hard look at the AI tools you're paying for and ask: Is this tool acting as a bicycle or a couch? If it’s automating a repetitive administrative chore, let it run. But if it’s touching clinical nuance, value-based contracting, or patient empathy, pull back and inject some deliberate human friction.
The goal of digital health innovation isn't to put our industry on autopilot. It's to handle the noise so that we can operate at the absolute ceiling of our ability. By drawing a clear line where the machine ends and human accountability begins, we don't just protect our skills, we build a sharper, smarter organization led by human ingenuity, calorie-burning brain cells, and spirited competition.
