AI sees a tumor. A good doctor sees you.

A Harvard study just dropped a number that should make us all sit up: AI diagnostic tools now match or exceed emergency room physicians in accuracy. We're talking about complex cases, real-time decision-making, the messy triage scenarios where seconds matter and information is incomplete.

And yet.

If you wake up tomorrow with chest pain, you're not going to want an algorithm. You're going to want a doctor. A human being who looks you in the eye and says "Let's figure this out."

That instinct isn't nostalgia. It's not technophobia. It's something deeper, and it's worth understanding why.

AI can read your scans. It can't read you.

Here's what AI does brilliantly: pattern recognition at inhuman scale. Feed it ten thousand mammograms and it learns what suspicious calcifications look like faster than any radiologist who's been reading films for twenty years. It doesn't get tired at 3 PM. It doesn't have a bad day. It just processes and flags and moves to the next image.

But here's what it can't do.

It can't see the way your hands shake when you're waiting for results. It can't hear the tremor in your voice when you ask "Is it serious?" It can't pick up on the fact that you're minimizing symptoms because you're terrified of what treatment might mean for your job, your kids, your life.

Joanna Stern from the Wall Street Journal recently wrote about letting AI analyze her breast imaging. The technology worked exactly as promised. It flagged an anomaly. It was accurate. And she was still glad when a human doctor sat down with her afterward to explain what it meant, what came next, what her options were. The AI gave her data. The doctor gave her a conversation.

That difference is everything.

Medical care happens to whole people, not isolated body parts. You're not just a set of symptoms to decode. You're someone's parent, someone's partner, someone with a mortgage and a vacation planned next month and a deep fear of needles you've never quite gotten over. AI sees the tumor. A good doctor sees you, scared and trying to hold it together, and adjusts how they deliver information accordingly.

This isn't a bug in human psychology - it's a feature

The instinct to want another human in the room when you're vulnerable isn't irrational. It's wired into us for good reason.

We evolved to read faces, to trust people who meet our eyes, to feel safer when someone who knows our name is making decisions about our body. That's not weakness. That's survival instinct playing out in a modern context.

Here's the thing people miss when they talk about AI in medicine: medical decisions aren't purely technical problems. Yes, diagnosis requires pattern matching and data synthesis. But treatment requires weighing tradeoffs, understanding what matters to this specific patient, navigating uncertainty in ways that involve values, not just variables.

Should we pursue aggressive treatment or focus on quality of life? How much risk is acceptable for a potential benefit? What does 'success' even mean when you're 82 versus 42?

AI can tell you what the data says. It cannot tell you what the data means for your life. It cannot hold the weight of the choice with you.

The 'humans want humans' instinct isn't holding medicine back. It's protecting us from the fantasy that healthcare is just an optimization problem. It's not. It's a deeply human thing happening to deeply human people.

The best future isn't AI replacing doctors: it's AI freeing doctors to be more human

So what does AI actually do for medicine if it's not replacing doctors?

It takes away routine basic analysis and gives them time to focus on the complex cases where they are really needed.

Right now, your doctor spends enormous chunks of their day on pattern-matching tasks that AI genuinely does better. Reading routine scans. Flagging lab abnormalities. Sorting through symptoms to generate a differential diagnosis. All critical work. All exhausting. All taking time away from what humans actually need from their doctors.

When AI handles that cognitive grunt work, doctors get to do what they trained for: the human parts. Listening to your story. Explaining complex information in ways that land. Noticing the question you're not asking but clearly want to. Sitting with you in the uncertainty.

The doctor's role doesn't disappear. It shifts. From diagnostic machine to guide. From gatekeeper of information to interpreter and advocate. From the person who spots the problem to the person who helps you figure out what to do about it.

This isn't a downgrade. This is an upgrade.

You get better diagnosis and more human attention. The AI catches what the tired resident might miss at hour twelve of their shift. The doctor gets to actually talk to you about what that means, instead of rushing to the next patient because there's a stack of charts waiting.

That's not compromise. That's both things working together the way they should.

AI will keep getting better at diagnosis. That's wonderful. That's going to save lives.

If you come into the ER with a broken bone, today's AI will certainly pick that up. But that same AI might miss a subtle, nuanced, post-op anomaly an experienced human doctor has seen before or read about. AI is really good at detecting the things it was trained for, but it hasn't seen everything. It hasn't met you.

Think of AI in medicine as just another Large Language Model (LLM) trained on a giant population of data. Because it is based on everyone it is not tailored to any one individual. At the end the day, AI is still simply processing billions of numbers and, to it, you are just another one.

Because of that, AI will never make your doctor obsolete, because medicine isn't just about being right. It's about being human when it matters most. It's about someone looking you in the eye when you're scared and saying "We'll figure this out together."

No algorithm can do that. And honestly? We shouldn't want it to.

Steve Chazin makes AI make sense. After three decades leading tech teams at companies like Apple and Salesforce, he's on a mission to show regular people how to use AI without fear or confusion. Welcome to #AIForTheRestofUs. stevechazin.com

I want to thank my son Jonah Chazin, MD for his valuable insights on this article. I appreciate all you do for your patients at Dartmouth-Hitchcock Medical Center and for all you do for your family every day. I love you, Jonah.