Recently, I was asked to sit on a faculty panel about artificial intelligence (AI) with computer science professors at Carnegie Mellon University — a leading institution for AI and robotics development. One of the panelists made a claim that I hear time and again in settings like this: “AI is just a tool.”
The implication is that AI is neutral, neither good nor bad. It’s up to us to use it well. An axe, for example, can cut wood for the family hearth, or can be used as a murder weapon.
This is simply not the case.
I have spent much of my career studying the financing and organization of the American health care system. I also co-own a small primary care practice with my wife. I have witnessed a rapid proliferation of AI use in American health care. Several tools — often referred to as AI-enabled Clinical Decision Support Systems (CDSS) — have recently been designed to help clinicians make diagnoses and develop treatment plans. OpenEvidence is arguably the best known, and recent reports suggest that 65% of all physicians are using it.
When we consider adopting any technology in a health care setting, we tend to focus on what it does — how it makes a given action easier. These AI-enabled systems are meant to address real issues. Medical evidence proliferates quickly, and it can be hard for clinicians to keep up. AI helps clinicians comb through evidence to properly diagnose, develop treatment plans — and do it all very efficiently. Clinicians are already rushed. The average visit time in primary care is approximately 15 minutes. Ostensibly, OpenEvidence allows for a more efficient encounter — allowing the clinicians to reach a diagnosis and develop a treatment plan more quickly.
We often focus less, however, on how technology impacts clinicians and, by extension, the broader healthcare system, where these technologies are even more deeply embedded.
One impact that I often worry about is that clinicians will lose important capabilities as they use AI. While such technologies allow clinicians to make decisions more quickly and — perhaps more accurately — there is certainly a sort of cognitive offload that happens. As they rely on these tools, clinicians may become cognitively weaker, and less equipped to make these clinical judgments on their own.
The same concern applies to several aspects of everyday life. ChatGPT or Claude will help your child finish schoolwork quickly, but will they develop the same cognitive abilities and skills that their parents did?
The Catholic Church — most recently Pope Leo XIV, and before him, Pope Francis — have voiced a concern with technology use on a deeper level. Technology does not just weaken many of our capabilities but shapes what we value.
“We have to accept that technological products are not neutral, for they create a framework that ends up conditioning lifestyles and shaping social possibilities,” argued Pope Francis in his 2015 encyclical “Laudato Si' ” (“Praise Be to You”). “Decisions that may seem purely instrumental are, in reality, decisions about the kind of society we want to build.”
In other words, as we use digital technology, we are conditioned by this “framework” — guiding our decisions about how to shape our world. Francis called this framework the “technological paradigm.”
Leo invoked this same theme in “Magnifica Humanitas” (“Magnificent Humanity”). He describes the technological paradigm as “the tendency to let the logic of efficiency, control, and profit alone shape personal, social, and economic decisions.”
And when that paradigm begins to dictate what matters and what can be discarded, it “reduc[es] creation to an object of exploitation and human beings to mere cogs in a system driven toward ever greater efficiency.”
What does this mean in practice? If the popes are correct, the technological paradigm guides the design and deployment of AI. One of its central features is the dogged pursuit of efficiency. Efficiency itself is not inherently bad, but it is an “instrumental” good — meaning that it is not good for its own sake but must be in service to a more important good.
The concern is that in healthcare, when engaging patients through tools designed for efficiency, clinicians may begin to see the clinical encounter itself as something that should be done quickly. The patient is no longer seen as a person worthy of time and attention but a problem to be rapidly solved and dispensed with — moving on to the next problem to be solved. The patient becomes a cog in this efficiency machine.
Even if clinicians themselves are not implicitly viewing the patient in this way, their decision to use these tools likely contributes to developing and reinforcing a broader system oriented toward mere efficiency. Along with Leo and Francis, even St. Pope John Paul II raised a big concern with systems ruled by an efficiency obsession: its tie to profits.
“The Church acknowledges the legitimate role of profit as an indication that a business is functioning well,” wrote John Paul. “But profitability is not the only indicator of a firm’s condition.”
Such a technological paradigm means that efficiency gains do not always serve the clinicians or patients, who are seen as a means to the end of profit maximization.
Both patients and physicians have long complained that they feel constantly rushed in their encounters, and that the health care system is overly oriented to profits. The proliferation of AI, though meant to improve the health care system, may make these problems even worse.
