The Power of Perspective: What AI Can Teach Us About Seeing the Whole Patien
A Reflection on Missed Diagnoses, Human Limits, and the Role of Technology in Restoring Context
Not long ago, I came across a story that stayed with me. A young boy, Alex, had spent three years visiting seventeen different specialists. Despite the best intentions of his providers—pediatricians, dentists, orthodontists, and more—none of them were able to piece together a clear diagnosis. It wasn’t until his mother, out of sheer frustration and love, turned to ChatGPT that the suggestion of tethered cord syndrome emerged. It made sense. And it changed everything.
What struck me wasn’t just the diagnostic miss. It was the pattern. Each clinician had a narrow view—accurate, perhaps, but constrained. Each diagnosis made sense in its own domain, but no one was stepping back far enough to see the whole picture. The dentist focused on jaw alignment, the pediatrician on milestones, the orthodontist on skeletal formation. Each was holding a piece of the puzzle, but no one was invited—or able—to assemble it.
“We kept hearing parts of the answer,” Alex’s mother said, “but nobody was willing to solve for the greater problem.”
That resonates more broadly than we’d like to admit. Our system rewards precision in the parts, but not perspective across them. Specialization has made remarkable advances possible, but it’s also made it harder to see what binds them together.
In my own clinical practice, I’ve seen the same fragmentation play out in different forms. Subspecialists working in silos. Care plans that conflict. Missed diagnoses not because of incompetence, but because of limited vantage points. And while collaboration exists in principle, it’s often undermined by time constraints, electronic medical record silos, and reimbursement models that reward volume over synthesis.
What happened in Alex’s case is remarkable, not because it was dramatic, but because it was so common. But what’s new—and worth reflecting on—is the role that AI played. ChatGPT was able to integrate symptoms and test results across domains and offer a plausible diagnosis—not because it was smarter than any individual clinician, but because it didn’t share the same blind spots.
Humans, for all our empathy and depth, carry cognitive biases. We over-index on our expertise, or the last case we saw, or the discipline we trained in. We’re rushed. We’re overwhelmed. In contrast, AI doesn’t carry these constraints. It has no stake in a single field or approach. It sees connections based on pattern, not pedigree. And in that lies both its strength and its limitation.
We shouldn’t see AI as a replacement for clinical judgment. It’s not. But it can be an ally—a way to hold a mirror up to our thinking and ask whether we’ve missed something. In diagnostic challenges like Alex’s, that capacity could be lifesaving.
But implementation matters. We must resist the temptation to over-rely or blindly adopt. AI’s utility depends entirely on how it’s designed, trained, and governed. It must be transparent. Auditable. Responsive to feedback. As Dr. Andrew Beam at Harvard noted, “AI tools like ChatGPT may not have the same blind spots as the human physician has.” That’s promising—but it’s not the same as saying it sees clearly either. It sees differently. And that’s precisely what we need in a field overwhelmed by complexity.
More than 12 million Americans are affected by diagnostic errors each year, and an estimated one-third of those errors result in serious harm. We know that rare and complex conditions are among the most common causes of missed diagnoses. These are precisely the kinds of situations where AI might help—not by replacing the clinician, but by surfacing possibilities that deserve a second look.
What’s needed now is an integrated system that recognizes both the strengths and the limitations of each contributor—human and machine alike. We can’t continue building care systems that only reward subspecialized knowledge. We need to create space for synthesis—for stepping back. That’s what AI offers us: the potential to augment our perspective, not override it.
If we’re thoughtful, AI can help restore what’s been missing in medicine: the ability to see the whole person, not just the part they happen to present with that day. It can help us reconnect the pieces we’ve allowed to drift apart. Not by dictating the answer, but by reminding us to look again. And sometimes, that second look is everything.
We owe our patients nothing less.
Dr. Salim Afshr