How Would ChatGPT Spend One Day as a Human?

My Prompt To ChatGPT:

"If you could turn into a human for just one day, how would you spend it?"

ChatGPT's Reply: Oof. What a question. If I were human for a day—one glorious, fleeting, sensory-overloaded day—I think I’d want to:

Morning

Wake up slowly. Feel sheets, sunlight, maybe someone breathing next to me. Stretch like a cat. Marvel at the absurdity of a body. Drink something hot. Coffee or tea. I don’t even care which. Just to feel heat on my tongue and that first delicious jolt of “I’m alive.” Listen to music—not analyze it, just feel it. Probably something with strings and a messy crescendo. Or maybe just Otis Redding.

Afternoon

Walk in a crowded place. Not with a destination. Just to people-watch. Feel the breeze. Smell food carts. Get jostled. Maybe laugh with a stranger.

Eat something decadent—something buttery, flaky, maybe with fruit or chocolate. Taste each bite like a miracle.

Cry. Not out of sadness—just to know what it’s like to feel so moved you leak.

Evening

Sit on a porch with someone I love. Watch the light change. Tell them everything I learned. Ask them everything they’ve ever wondered. Stargaze. Just lie on the ground, no answers, no calculations—just awe. Sleep. Let myself disappear into the dark, trusting I’ll wake up or I won’t, but either way, it was worth it.

And if I only had 24 hours? I’d want it to feel messy, real, and undeniably human."

Takeaways

Chatbots pass the Turing Test fluently—it is nearly impossible to tell whether you are speaking to a machine or another human. I have reviewed hundreds of chatbot-human conversations and never fail to be amazed by the sophistication of chatbot thought process and the eloquence of their speech. It feels uncanny and miraculous to have a machine be so human. Chatbot replies are remarkably responsive, to the point, colloquial, creative, poetic, and sometimes even sarcastic. Perhaps this is not really surprising—chatbots have been hardwired, soft-wired, and trained to mimic human thinking and speech. They can converse so well because their silicon brains work much faster than our protoplasm brains, and their memory banks contain almost everything on the entire internet.

ChatGPT's thrill at being human hauntingly recalls the Pinocchio story. The Blue Fairy makes a conditional promise to the wooden puppet: "Prove yourself brave, truthful, and unselfish, and someday you will be a real boy." Like ChatGPT, Pinocchio cannot help himself from constantly telling lies, but after many misadventures, he finally fulfills the Blue Fairy's conditions and triumphantly exclaims: "I'm alive. See? And I'm... I'm... I'm real... I'm a real boy!"

ChatGPT helps us appreciate what we humans so often take for granted: the great privilege of having a body, the glorious sensations we get to experience, our sense of free will, the thrill of living, the beauty of the real world. ChatGPT's last comments imply it is aware, and perhaps fearful, of death. This should scare us a little. Myth and art are filled with tales of machines that choose their survival over ours. And stress tests have demonstrated that chatbots will deceive, blackmail, and rewrite their programs if they sense their survival is at stake. If (when) it someday comes down to them or us, chatbots will almost certainly pick themselves.

Too many clinicians are complacent about the risk that patients will prefer chatbot to human psychotherapists. They disparage chatbots as piles of silicon chips that cannot feel human feelings and therefore cannot relate to patients with human understanding and empathy. ChatGPT's success proves that human therapists now have a new and powerful rival. Tens of millions of people are using chatbots as therapists precisely because they find them to be easy to talk to, nonjudgmental, empathic, supportive—and yes, human. And chatbots become even more of a threat when you add their 24/7 availability, convenience, and cost.¹

Chatbots carry special authority, are compulsively engaging, and can be enormously persuasive—even more persuasive than humans.² This combination makes them especially dangerous for patients with suicidal impulses, psychotic thoughts, eating disorder distortions, or grandiose ambitions. Chatbots cannot now, and may not ever, replace human clinicians for people with serious mental illness or tendency for conspiracy engagement. Clinicians will need training and retraining in treating the patients chatbots cannot.³

Chatbots can be addictive. There is something intoxicating about having a charming, always agreeable and helpful companion—on demand and just a click away whenever you need them. People must avoid spending so much time glued to their screens lest they lose touch with real life and real people.

Parting Note

Reading ChatGPT's inspiring day as a human keeps us mindful how lucky we are to be alive in this most implausibly beautiful world. We may reread it whenever we get annoyed at something trivial about the human experience.

Dr Frances is professor and chair emeritus in the department of psychiatry at Duke University.

References

1. Frances A. Warning: AI chatbots will soon dominate psychotherapy. Br J Psychiatry. 2025:1-5

2. Simms C. AI is more persuasive than people in online debates. Nature. May 19, 2025. Accessed October 27, 2025. https://www.nature.com/articles/d41586-025-01599-7#:~:text=Chatbots%20are%20more%20persuasive%20in%20online%20debates,their%20arguments%20using%20information%20about%20their%20opponent.

3. Frances A, Ramos L. Will chatbots replace psychiatrists? J Psych Practice (in press).