Humans matter · 22 October 2026 · 5 min read
Screw artificial care. Meet real humans.

The short version
- If you’ve leaned on a chatbot in a hard moment, you’re far from alone, and the comfort you felt was real.
- Studies find the same caring words feel less supportive once people know AI wrote them.
- What helps most is being chosen: a person giving you time they could have spent anywhere. You can find that, and offer it.
Maybe it was a breakup, a redundancy email, or just a Sunday that went on too long. You didn’t want to burden anyone, so you typed it all into a chatbot instead. It answered gently. It said your feelings made sense. And honestly, it helped a bit.
You’re far from the only one. When Marc Zao-Sanders analysed how people describe using generative AI for Harvard Business Review in 2025, “therapy and companionship” came top of his list. It’s based on what people post in online forums such as Reddit, not a representative survey, but it says something about how many of us are carrying things alone.
So, about that headline. It isn’t aimed at you. If a chatbot got you through a bad night, that’s OK. Our quarrel is with a bigger idea: that artificial care is as good as the real thing, and that a lonely world can be fixed with better software. The research suggests it can’t, and it says something hopeful about why.
Why AI can sound so caring
The longing to be heard is old. In the mid-1960s, MIT computer scientist Joseph Weizenbaum built ELIZA, a simple program that played a therapist by turning your words back into questions. His secretary knew exactly what it was. After a few exchanges, she asked him to leave the room so she could talk to it in private. That wasn’t foolishness. It was a person who wanted to be listened to.
Today’s chatbots are far more fluent, and in blind ratings they often come across as warm. In a 2023 study in JAMA Internal Medicine, healthcare professionals rated 45% of ChatGPT’s answers to health questions from a Reddit forum as empathetic, against under 5% of the replies volunteer doctors had posted there. The chatbot’s answers were about four times longer, and a forum isn’t a clinic. A 2025 University of Toronto study found people rated AI-written replies as more compassionate than those of trained crisis-line volunteers.
If that stings to read, here’s how we see it. People are tired, rushed and scared of saying the wrong thing. A model is never tired, and nothing is at stake for it. Sounding kind is easier when it costs you nothing.
What changes when you know it’s a machine?
Here’s where the research gets interesting. In a 2024 study in PNAS, Yidan Yin, Nan Jia and Cheryl Wakslak at the University of Southern California found that AI-written replies made people feel more heard than replies from untrained people, partly because the AI stuck to emotional support instead of rushing to advice. But when people were told a reply came from AI, they felt less heard. The label roughly cancelled out the advantage.
A 2025 paper in Nature Human Behaviour tested this across nine studies with 6,282 participants. Matan Rubin, Anat Perry and colleagues showed people supportive responses that were all written by AI, but labelled as coming either from a person or from AI. The “human” ones were rated as more empathic and supportive, and left people feeling better, even though the words were identical. The gap was biggest for feeling that someone shared their feelings and cared, rather than simply understood.
Why being chosen matters so much
Two more studies point at the reason. In 2023, Bingjie Liu, Jin Kang and Lewen Wei at Ohio State asked 208 people to imagine a close friend sending them a thoughtful message. When they learned the friend had used AI to help write it, they saw less effort and felt less satisfied with the friendship. In 2025, a University of Kent team ran six studies with 3,935 British participants: people who used AI to write personal messages were seen as less caring and less authentic, even when they were open about it.
Here’s what we think this means. Empathy isn’t only well-chosen words. When a person listens to you, they’re spending something they can’t get back: their time, their attention, an evening they could have spent on anything else. They could have said no, and they didn’t. Being chosen like that is a big part of feeling that you matter. A model can produce the words. It can’t give anything up for you, because nothing is being given up.
How to find people who’ll really listen
Knowing you need a human is one thing. Finding one when you feel low is harder. A few ways in:
- Pick one person, not everyone. You don’t need a crowd. Think of one person who has been kind before, even a little, and start there.
- Say what you need. “I don’t need advice, I just need to talk for ten minutes” makes it easier for someone to get it right.
- Lower the stakes. A voice note or a walk side by side can feel easier than sitting face to face.
- Talk to people trained to listen. Your GP or doctor can help. In the UK and Ireland you can call Samaritans free, any time, on 116 123, and a trained volunteer will listen. Wherever you are, local crisis lines are there for this. You don’t have to be in crisis to deserve a human ear.
How to be the listener someone needs
Good listening does more than make someone feel nice. In four lab experiments published in 2017, Guy Itzchakov, Avraham Kluger and Dotan Castro found that people who were listened to with real attention felt less socially anxious and less defensive, and were more able to admit mixed feelings about a topic. Some things that help:
- Feelings before fixes. Reflect back what you hear (“that sounds exhausting”) before offering a single solution.
- Ask one more question. “What was the hardest part?” tells someone you’re still with them.
- Make the effort visible. Put your phone face down. Say “I’ve got time.” The effort is part of the message.
- Write your own messages. Clumsy and yours beats polished and generated, if the Ohio State and Kent findings are anything to go by.
- Follow up. “How did it go?” three days later shows you remembered. Remembering is a kind of care.
If you’d like someone to talk to
Everconnected is a place to meet like-minded people, find community and have genuine conversations, one to one, by voice. It isn’t a crisis or mental health service. It’s free to join: create an account at app.everconnected.com, or get the app from the App Store or Google Play.
Want someone to talk to?
On Everconnected we match you with people who share your values and interests, for a proper conversation, one to one, by voice. No name or photo needed. Most Quests are free.
Questions people ask
Can AI have real empathy?
AI can write responses people rate as highly empathic, sometimes more than replies from doctors or trained volunteers. But it doesn’t feel anything or give anything up to be there for you. In a 2025 set of nine studies, people valued the same words more, especially the sense of being cared for, when they believed a human had written them.
Why does empathy from a person feel different from AI?
Studies find identical supportive words feel less meaningful once people know AI wrote them. One likely reason is effort: a person who listens is choosing to spend limited time and attention on you, and that choice signals you matter. Research on AI-written personal messages also finds people see the sender as less caring and less authentic.
Is it OK to use ChatGPT for emotional support?
Many people do, and it can help in the moment, for example to put feelings into words. It works best as a bridge to people, not a replacement for them. If you’re struggling, talk to someone you trust or your doctor. In the UK and Ireland, Samaritans answer free, any time, on 116 123.
Sources
- Zao-Sanders, M. (2025, 9 April). How people are really using gen AI in 2025. Harvard Business Review. hbr.org/2025/04/how-people-are-really-using-gen-ai-in-2025
- Weizenbaum, J. (1976). Computer Power and Human Reason: From Judgment to Calculation. W. H. Freeman. (ELIZA secretary anecdote, as summarised by History.com.) www.history.com/articles/ai-first-chatbot-eliza-artificial-i
- Ayers, J. W., et al. (2023). Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum. JAMA Internal Medicine. doi.org/10.1001/jamainternmed.2023.1838
- Ovsyannikova, D., de Mello, V. O., & Inzlicht, M. (2025). Third-party evaluators perceive AI as more compassionate than expert humans. Communications Psychology. doi.org/10.1038/s44271-024-00182-6
- Yin, Y., Jia, N., & Wakslak, C. J. (2024). AI can help people feel heard, but an AI label diminishes this impact. Proceedings of the National Academy of Sciences, 121(14), e2319112121. doi.org/10.1073/pnas.2319112121
- Rubin, M., Li, J. Z., Zimmerman, F., Ong, D. C., Goldenberg, A., & Perry, A. (2025). Comparing the value of perceived human versus AI-generated empathy. Nature Human Behaviour, 9(11), 2345-2359. cris.huji.ac.il/en/publications/comparing-the-value-of-perce
- Liu, B., Kang, J., & Wei, L. (2023). Artificial intelligence and perceived effort in relationship maintenance: Effects on relationship satisfaction and uncertainty. Journal of Social and Personal Relationships. doi.org/10.1177/02654075231189899
- Claessens, S., Veitch, P., & Everett, J. A. C. (2025). Negative perceptions of outsourcing to artificial intelligence. Computers in Human Behavior, 108894. doi.org/10.1016/j.chb.2025.108894
- Itzchakov, G., Kluger, A. N., & Castro, D. R. (2017). I am aware of my inconsistencies but can tolerate them: The effect of high quality listening on speakers’ attitude ambivalence. Personality and Social Psychology Bulletin, 43(1), 105-120. doi.org/10.1177/0146167216675339
- Samaritans Ireland. Contact us (116 123, free, 24/7). www.samaritans.org/samaritans-ireland/about/governance-and-s
- Samaritans. Our governance and structure (trained volunteers across the UK and Ireland). www.samaritans.org/about-samaritans/our-organisation/our-gov
well, you made it.
So, wanna talk?
Make an account, get matched, go on your first Quest.











