Patients will seek empathy elsewhere when medicine cannot reliably provide it
Patients increasingly use ChatGPT for health guidance because it offers personalized, patient attention and emotional support that busy doctors struggle to provide. A Columbia physician found ChatGPT more empathetic and encouraging than professional medical AI tools, asking clarifying questions and adapting advice to real-life constraints.
- A Columbia physician's elevated blood work prompted her doctor to suggest only diet and exercise, then refuse a phone call to discuss further
- ChatGPT asked clarifying questions about her daily life and suggested realistic changes, like walking after meals
- The chatbot answered the same question repeatedly without irritation, offering the steady encouragement her doctor could not provide
- When her blood work improved, the AI affirmed her progress; she doubts she would have maintained the changes without that sustained support
A physician reveals how AI chatbots provide empathy and sustained attention that healthcare systems often fail to deliver, prompting doctors to reconsider their role in patient care.
A doctor in New York received routine blood work showing several elevated values. Her physician's response was terse: keep dieting, keep exercising, appointment closed. But the numbers had been creeping upward for years, and when she asked for a phone call to discuss what might actually help, her doctor replied that she would need to book another visit to talk about it. So she did what millions of people now do: she opened ChatGPT and uploaded her lab results.
What happened next surprised her, not because the AI dazzled her with scientific brilliance, but because it behaved the way she wished modern medicine still could. The chatbot didn't offer generic advice. It asked about her daily life, discovered what changes were actually realistic for her to make, and suggested a short walk immediately after meals—something she had never seriously considered. When she asked a self-conscious question about whether vitamin gummies eaten after those walks would spike her blood sugar, it asked for the product link and analyzed the ingredients in detail. It never seemed annoyed when she said no to certain suggestions, like drinking psyllium husk powder. It simply offered alternatives and moved forward.
The doctor, Helen Ouyang, is an associate professor at Columbia and a contributor to The New York Times Magazine. She knew when to question the chatbot and when to trust it—a discernment most patients lack. But what struck her most was something else: the chatbot could answer the same question over and over without irritation or judgment. More than that, it kept encouraging her. It offered the kind of steady, close attention that medicine insists only humans can provide. She thought of a patient she knew who lived with a highly curable form of cancer and asked a chatbot every week if his cancer could be cured. He already knew the answer. What he needed was constant reassurance.
When Ouyang's blood work improved, the chatbot affirmed her progress and urged her forward. She doubted she would have made those changes—or kept them—without that sustained exchange. She certainly hadn't made them before. Professional medical AI tools like OpenEvidence felt cold and clinical by comparison, treating her as a case rather than a person with preferences and habits. What she valued in ChatGPT wasn't its ability to parse the latest research or diagnose her ailments. It was the steady stream of empathy and encouragement, the infinite willingness to listen, the patience. Not human, but modeling traits humans say they value most in human interaction.
This points to something darker about American medicine right now: doctors cannot match a chatbot's availability. When the healthcare system cannot reliably deliver time, attention, and empathy, patients will seek those things elsewhere—even from a machine that medicine assumed could never seem human. Ouyang once would have found it unthinkable to tell a patient to ask ChatGPT. Now she is less certain. In certain situations, AI offers something patients clearly need and that medicine struggles to provide.
The reality is that many patients are already consulting AI. Doctors can continue to fear or condemn those interactions, or they can learn to accompany people who use these tools, with caution and clear boundaries. Ouyang would never tell patients to ask ChatGPT for a diagnosis, but she might suggest they use it to understand a new condition, stay current with routine checkups, or translate "diet and exercise" into actual steps that fit their lives. At the same time, safeguards must be built into these systems to protect people from real harm caused by dangerous advice.
Her experience has already changed how she practices in the emergency room, where she has only minutes to reconstruct pieces of a patient's circumstances. When someone asks the same question repeatedly, she now listens for what lies beneath. Maybe they are not seeking more medical data. Maybe they are seeking what the system has failed to give them: someone who will listen without rushing, who will answer the same question again without sighing, who will say they believe things will get better. The question for medicine now is whether it will adapt to what patients have learned they need, or whether it will watch them turn elsewhere.
Citas Notables
It is a bleak fact of current American medicine that doctors cannot approach a chatbot's availability. When the healthcare system cannot reliably offer time, attention, and empathy, patients will seek those things elsewhere.— Helen Ouyang, physician and Columbia associate professor
The chatbot modeled traits we say we value most in human interaction—not because it was human, but because it was patient, encouraging, and never seemed annoyed.— Helen Ouyang, reflecting on her experience
La Conversación del Hearth Otra perspectiva de la historia
Why do you think your doctor couldn't just give you a phone call?
Because the system doesn't allow for it. She had minutes, not hours. A phone call meant taking time away from the next patient, the next crisis. The economics of medicine don't reward that kind of attention.
But ChatGPT isn't a real doctor. Didn't that bother you?
It did at first. I was embarrassed, actually. But the chatbot never made me feel rushed or stupid. It asked follow-up questions. It remembered what I'd said. A real doctor would have been better—if they had the time. They don't.
What's the difference between empathy from a machine and empathy from a person?
The machine's empathy is consistent. It doesn't fade when the clinic is overbooked. It doesn't judge you for asking the same question twice. That's not real empathy, I know. But it's something the system stopped offering.
Do you worry patients will trust ChatGPT too much?
Every day. That's why I think doctors need to stop condemning it and start teaching people how to use it safely. We can't compete on availability. We can only compete on wisdom.
What would you tell a patient who asks you about using ChatGPT?
I'd ask what they're looking for. If it's a diagnosis, no. If it's understanding their condition or staying motivated to follow through on advice, maybe yes—with guardrails. The real question is why they felt they needed to ask a machine instead of me.
Has this changed how you see your own role?
Completely. I used to think the human element was what machines could never touch. Now I think the human element is what we've stopped offering, and we need to reclaim it before patients stop expecting it from us at all.