Contrary to Hollywood portrayals of rogue machines, today’s AI risks are far less dramatic—but potentially far more expensive. 

The greatest risks are not machines replacing clinicians; they are widening health disparities, increasing healthcare costs, and expanding the environmental footprint of increasingly energy- and water-intensive data centers.

More importantly, many of healthcare’s biggest challenges are not technological. Inefficient workflows, fragmented health records, poor interoperability, opaque payment models, and unequal access to care are fundamentally human and policy problems. Layering AI onto broken systems without first addressing these foundational issues simply allows expensive inefficiencies to run faster.

Digital therapeutics perfectly illustrate both the promise and the limitations of AI-enabled healthcare. The U.S. Food and Drug Administration has now cleared prescription digital therapeutics for conditions such as generalized anxiety disorder, representing an important milestone for evidence-based digital medicine. 

These products demonstrate that software, when rigorously evaluated, can become a legitimate therapeutic intervention.

Unfortunately, many of these innovations are available primarily through employers or commercial health insurance plans. Since approximately 27 million Americans remain uninsured, many of the people who could benefit the most from these technologies may never gain access (CDC, 2025).

It’s a little like inventing the world’s safest self-driving car and then announcing that only people who already own a luxury vehicle are allowed to drive it.

The same concern applies globally. Deploying increasingly sophisticated AI on top of healthcare systems that still cannot exchange information effectively is like installing a Formula 1 engine into a shopping cart. It certainly becomes faster—but not necessarily safer, more efficient, or capable of reaching its destination.

The environmental implications deserve equal attention. Modern generative AI models require enormous computational resources, consuming substantial electricity and water to train and operate. As healthcare organizations rapidly adopt AI, leaders must ask an important question: Are we creating sufficient clinical value to justify the environmental and financial resources being consumed?

Healthcare has always operated under finite resources. Every dollar invested in AI should be done bearing in mind the need for  preventive care, workforce development, primary care access, interoperability, or addressing the social determinants of health.

Responsible AI adoption therefore requires more than technical excellence—it requires thoughtful prioritization.

Technology alone rarely fixes broken systems. History has repeatedly shown that digitizing inefficient processes simply creates faster inefficiency. AI is no exception.