There is a version of medicine's AI future that Russell Kennedy finds genuinely exciting. A physician can query thousands of research papers in seconds, catch a pattern in an imaging scan that human eyes might miss, and deliver a more accurate diagnosis faster than was possible a generation ago. In Kennedy's framing, AI is "a remarkable extension of the analytical mind," and he means that as high praise. The efficiency gains are real. The reduction in errors is real. The democratization of high-quality medical knowledge, for both clinicians and patients, is real.
But Kennedy's specialty shapes what he sees when he looks further down that road. As a physician focused on anxiety disorders and emotional dysregulation, he has spent his career watching what actually precedes recovery in his patients. What he has observed is not a function of information transfer. People begin to heal, he says, when they feel "safe, seen, heard, understood, and genuinely cared for." That process is biological as much as it is psychological. The social engagement system depends on eye contact, tone of voice, facial expression, body language, and what Kennedy describes as "the subtle sense that another human being is truly present with us."
That is where his concern with AI sharpens into something specific. The worry is less that AI will make a wrong diagnosis and more that medicine will quietly drift toward treating illness as, in his words, "merely a problem of logic and information." The tools get better, the models grow more sophisticated, and somewhere in that progression the field forgets that human beings are not machines.
"AI may be able to simulate empathy with increasingly convincing language, but it does not actually feel. And there is something profoundly important about being with another human who can."
Kennedy is not arguing against AI in clinical settings. His position is more precise than that. The question he is raising is whether the conversation happening at the policy and technology level is accounting for what he witnesses at the level of the nervous system. An AI system can generate warm, reassuring language. It can mirror the cadences of compassion. What it cannot do is be present in the way that registers, beneath conscious awareness, as genuine human contact. For patients whose nervous systems are already dysregulated by anxiety, that distinction may not be incidental to their care. It may be central to it.
The future Kennedy describes is one where computational intelligence and human connection are combined rather than traded against each other. What remains unresolved, and what his perspective keeps returning to, is whether the institutions designing and deploying these tools share that premise, or whether efficiency, scale, and the seductive precision of data will quietly crowd out the part of medicine that has never been measurable.
