What Texting Someone Cannot Do for You

You texted someone back within the hour. You liked their photo. You sent the meme you knew they’d laugh at. It felt like contact, and in a narrow sense it was. The question is what it was standing in for, and what it wasn’t.

This matters more than it used to, because a growing amount of what people count as their social life now happens through a screen with nobody else’s body anywhere near it — texting, social feeds, and increasingly, conversation with an AI companion. The American Psychological Association’s Monitor on Psychology reviewed the current evidence on AI chatbots and digital companionship in January 2026 and found something less alarming than the panic and less reassuring than the marketing: effects depend heavily on who’s using them, how the chatbot is designed, and how much time gets poured in. It’s not a simple story of harm or help.

The distinction that keeps getting collapsed

Loneliness and social isolation are not the same thing, even though people use them interchangeably. Isolation is a fact about your circumstances — how many people you have contact with, how often. Loneliness is a feeling — the gap between the connection you have and the connection you want. You can be surrounded by people and lonely. You can live alone and not be.

Online contact is unusually good at closing the isolation gap and unusually bad at closing the loneliness gap. A text thread means you are, technically, in touch with someone. It does the bookkeeping of a relationship: birthdays remembered, plans half-made, the sense that a person still exists in your life. What it does less well is deliver the thing that actually seems to protect health — and the evidence here is specific enough to be worth taking seriously rather than gesturing at.

What the body seems to register

Julianne Holt-Lunstad’s 2015 meta-analysis, covering over 300,000 people, found that social isolation, loneliness, and living alone each independently raised the odds of early death — isolation by 29%, loneliness by 26%. The American Heart Association’s 2022 scientific statement went further on mechanism, tying isolation and loneliness to roughly a 30% increased risk of heart attack or stroke, and to worse outcomes in people who already have cardiovascular disease. Nobody has run the equivalent trial asking whether an hour of texting a day changes those numbers, and the AHA statement itself says plainly that the intervention evidence — what actually reduces the risk, once you’ve identified it — is the field’s biggest gap. But the studies that do exist on connection and mortality were almost entirely built on in-person social contact: shared meals, live conversation, physical presence. That’s not a technicality. It’s the whole basis for the numbers.

A 2025 study of college students coming out of the pandemic, published in Frontiers in Psychology, is more directly useful here. It looked at which kinds of social connection actually predicted wellbeing once students were back around other people, and the pattern that emerged was familiar: quality and regularity of contact mattered far more than volume. A large network of people you occasionally message is not doing the same work as three people you see most weeks.

Where AI companionship actually sits

George Mason University’s public health commentary from late 2025 frames AI companions as a substitution risk rather than a straightforward remedy — the concern isn’t that talking to a chatbot is bad in itself, but that it’s easy, always available, and asks nothing of you, which makes it a comfortable place to stop rather than a bridge to somewhere else. The APA’s 2026 review is more agnostic: some people use AI companionship as a supplement while they build human contact elsewhere, and for them it may do no harm and even help with rehearsal — trying out a hard conversation before having it with a person. Others use it as a replacement, and for that group the evidence is worrying precisely because a chatbot never gets tired, never has its own bad day, and never needs anything back from you. That one-sidedness is comfortable. It is also, structurally, not the thing that protects your health.

What this changes about tonight

None of this means delete the group chat. It means noticing what job you’re asking it to do. A text thread is good at maintaining a relationship that already has some in-person weight underneath it — the friend you see every few weeks, kept warm between visits. It is much worse at creating that weight in the first place, and worse still at replacing it once it’s gone.

If your social contact this month has been almost entirely typed, the useful move isn’t a bigger phone habit or a better app. It’s converting one digital thread into one recurring, physical, scheduled thing — the same person, the same kind of Tuesday, ideally something that doesn’t depend on you both being in the mood that week. Eric Klinenberg’s research on social infrastructure makes the point that this kind of contact tends to need a place as much as an intention: a specific library, a specific run club, a specific table. Pick one relationship that currently lives mostly on a screen and give it a location and a repeat booking. The text thread can stay. It just isn’t the thing doing the protecting.