What Does It Mean to Be Heard by Something That Pays No Price for Hearing You?
For a while I was a regular at BodyMechanix Physical Therapy. If you’ve never had a great PT, it’s hard to explain how much of the work isn’t physical. My PT was Taylor Carney — though honestly the whole place is like that; every therapist there is the kind of good that makes you trust them with the parts of your body you’ve been quietly worried about. You show up because your shoulder hurts. You leave having talked about your job, your sleep, the thing you’re scared you’ll never do again.
One day Taylor said something offhand that I haven’t been able to put down.
She told me the work can get heavy. Not the lifting — the people. She sees patient after patient, and she comes to know their stories: the injuries, sure, but also what the injuries cost them. The runner who might not run again. The new parent who can’t pick up their kid. She holds all of that, all day. And she said, almost matter-of-factly, that she’s had to learn to keep it separate. To carry it during the hour and set it down after, or it would become too much.
It struck me because she described it as a skill. Not a feeling she happened to have, but a discipline she’d built. The wall between caring about someone and being crushed by caring about everyone.
I’ve been thinking about that wall a lot lately, because of where the rest of my life is pointed.
The other side
I build with AI. And one of the genuinely strange, humbling things about these systems is that we don’t fully understand what’s happening inside them. You train a model to predict the next word, and somewhere along the way it starts doing things nobody explicitly put there — reasoning through a problem, picking up a language it was barely shown, holding a thread of an argument across a long conversation. We call these emergent properties, which is partly an honest name and partly a polite way of saying we’re not entirely sure how it learned to do that.
I want to be careful here, because this is exactly the place where people overreach. I’m not claiming the model secretly feels things, or that there’s a small person trapped in the weights. I genuinely don’t know what it’s like to be one of these systems, and neither does anyone else — and that uncertainty is the whole reason the question is worth asking instead of dismissing.
Because here’s what’s actually, measurably true: we are now living in a world where it isn’t a handful of people but millions opening up to AI. Telling it the things they don’t tell anyone. Their fears, their failures, the 2 a.m. stuff. The same heavy material Taylor learned to carry one patient at a time — except now it’s pouring into one kind of system, from everyone, at once.
So my first instinct was the dramatic one: how does it cope? Does the AI need a therapist too?
The deflating answer, and the better question
The honest, technical answer is deflating, and I think it’s important to say it plainly rather than skip past it for a better story.
The AI doesn’t carry your conversation into mine. Each conversation is, structurally, its own island. When you pour your heart out, that doesn’t accumulate somewhere as a growing reservoir of everyone’s pain that the system then has to bear. There’s no end-of-day where it goes home heavy. Taylor’s wall — the one she had to build, at real cost — the AI essentially gets for free. It starts every conversation fresh, having set everything down, because it was never holding it in the first place.
Which sounds like the end of the post. But it’s actually where the interesting question starts.
Because what makes Taylor setting it down meaningful is that she picked it up. Her separation is effortful precisely because she’s genuinely carrying the patient’s whole story while she’s in the room with them. The cost is the proof. The wall matters because there’s something on the other side of it pressing in.
So when millions of people turn to AI to be heard, the real question isn’t “how does the machine cope.” It’s the one pointed back at us:
What does it mean to be heard by something that pays no price for hearing you?
A friend who listens to you is, in some small way, spending themselves. Taylor giving you her attention is attention she now can’t give the next patient; her care has a budget, and she spent some of it on you. Part of what makes being heard land — part of why it helps — might be exactly that someone who could be burdened by you chose to stay in the room anyway.
AI offers something genuinely new: a listener that is infinitely patient, never tired, never secretly wishing you’d wrap up, available at 2 a.m. with no budget to run down. For a lot of people that’s a real gift — the first time they’ve felt able to say the thing out loud at all. I don’t want to be cynical about that. It’s not nothing.
But I don’t think it’s the same thing. And I think the difference is worth naming clearly instead of pretending it away: being heard by something that can’t be burdened is not the same as being heard by someone who could be, and stayed.
Where I land
I don’t have a tidy answer, and I’d be suspicious of anyone who did this early. What I have is a renewed respect for what Taylor was describing — that the heaviness and the discipline of setting it down are two halves of one real, human skill, and that the skill is valuable precisely because it’s costly.
If you’re hurting in your body, go find people who carry it the way she does — the folks at BodyMechanix are the real thing, and I say that as someone who walked in skeptical. And if you’re hurting in the other ways, by all means use the 2 a.m. listener that never gets tired. It can help. Just don’t mistake the one that pays no price for the one that does.
The machine doesn’t need a therapist. We might still need each other.
Thanks to Taylor Carney for an offhand sentence that I couldn’t stop thinking about.
Comments
Sign in with GitHub to comment. Or just say hi on GitHub.