AI Will Outperform Doctors Clinically—But Can't Replace Presence

An interventional cardiologist who credits artificial intelligence with making him a dramatically better physician now warns that the technology's most dangerous impact may have nothing to do with diagnostic accuracy.

Andrey Espinoza, founder and CEO of the Advanced Heart and Vascular Institute of Hunterdon, acknowledges that AI already sharpens his differential diagnoses, catches subtle arrhythmia patterns, and frees him to spend more time with patients. He believes extrapolating current trajectories forward a decade or two leads to an honest conclusion: machines will likely outperform human physicians on technical grounds. Robotics already executes sub-millimeter precision beyond human capability, and pattern recognition systems already surpass experienced clinicians at reading angiograms and detecting rhythm changes.

But Espinoza draws a sharp line between technical superiority and the complete replacement of physicians, arguing the distinction rests on understanding that diagnosis is a fact while care is a relationship.

The unbillable work of medicine

The physician describes telling a patient about critical limb ischemia as two separate acts: delivering the clinical decision, which AI can handle, and sitting with the patient's fear and their family's fear—what he calls a "witnessing problem" rather than a processing problem. He contends that suffering wants to be seen by something that can also suffer, and that this forms the load-bearing wall of the doctor-patient relationship.

Espinoza recounts meeting families under the worst circumstances—introducing himself to a wife and two children in the same sentence he tells them their husband and father died from a massive heart attack despite all interventions. He describes going home afterward, watching his own children sleep, and feeling guilt that he gets to tuck them in on nights he just destroyed that same certainty for other families. That guilt, he notes, isn't billable. A machine can be right but cannot be haunted, and he believes good medicine requires the willingness to be haunted.

Why it matters

The structural threat Espinoza identifies goes beyond technology displacement. Healthcare reimbursement pays for procedures, not for staying in the room an extra ten minutes. Human presence has historically ridden along bundled with clinical necessity. If AI strips away the procedural work that has quietly subsidized the human work, the human work may simply stop being paid for—and therefore stop happening—long before any machine could actually replace it. The risk isn't machines outcompeting physicians at compassion, but healthcare systems quietly deciding compassion doesn't survive once it's no longer bundled with something billable.

Building for presence

Espinoza argues medicine needs to create structural and financial room for one human being to sit with another at the worst moment of their life, even after machines can make faster diagnoses and cleaner incisions. He believes the day physicians are no longer needed to practice medicine may arrive faster than most are prepared to admit, but that the day patients no longer need another human being in the room doesn't exist—and that medicine should be built as if it doesn't, on purpose, before the chance to choose is lost.

The essay was written by Andrey Espinoza, a triple board-certified interventional cardiologist and vascular specialist, and published on KevinMD.