By Jeanne Cohen

Artificial intelligence (AI) has quickly become the most talked-about technology in healthcare today. We are in the midst of the hype cycle, where promises and fears about AI are amplified to extremes. On one end of the spectrum, there are claims that AI will “save healthcare” by eliminating inefficiencies and solving decades-old challenges. On the other end, AI is portrayed as untrustworthy, opaque, and even dangerous.

The truth, as in most cases, lies somewhere in the middle. AI will not solve every problem in healthcare, nor is it inherently harmful. Instead, the real opportunity lies in hybrid intelligence, also referred to as human-in-the-loop, which is the deliberate and balanced combination of machine intelligence with human judgment. AI brings unparalleled speed, scalability, and analytic capabilities. Natural intelligence -- our clinical expertise, ethical frameworks, and ability to understand nuance -- adds the contextual wisdom that algorithms alone cannot provide. For healthcare, in particular, hybrid intelligence is not only the best path forward, but also the only sustainable one.
Why hybrid intelligence matters most in healthcare
Among all industries applying AI, healthcare has the most at stake. Patient outcomes, safety, and trust are all impacted by the technology and tools used in every aspect of healthcare. That’s why a hybrid intelligence approach needs to be the foundation for AI adoption in our industry. According to a recent research by Bain & Company patient comfort with AI is rising. The study showed that patients accepting AI (or AI scribes) for listening and taking clinical notes rose from 21% in 2024 to 60% in 2025. While the same study reported that only 34% of patients are comfortable with AI making a diagnosis and just 28% with AI becoming their doctor.

This hybrid model acknowledges both the strengths and limitations of AI. It leverages AI for what it does best -- rapid computation and pattern recognition -- while preserving the human elements of care that technology cannot replicate.

The imperative of explainability
One of the significant barriers to AI adoption in healthcare today is a lack of trust. Clinicians and administrators must believe that AI tools are reliable, transparent, and fair. That’s why explainability and the ability to see and understand the evidence, sources, and reasoning behind AI outputs is non-negotiable.

Explainability requirements should always correlate to the level of clinical risk. Let’s look at administrative tasks where basic traceability is enough to ensure processes are functioning correctly. AI tools that automate claims submission must be auditable but do not require complex clinical justification. In clinical decision support applications, however, there is much greater risk and therefore a requirement of transparency and validated sources such as professional society guidelines, peer-reviewed evidence, and clear oversight. AI outputs cannot be “black box” answers; they must be accompanied by evidence and open to expert review.

Artificial Intelligence Homepage