The Advice That Wants You Single

I saw something the other day that I haven’t been able to stop thinking about.

A bloke asked an AI chatbot for dating advice. He’d met a woman on an app, they’d had coffee, it went alright. She came back and suggested dinner. Even offered his birthday as one of the dates, which if you’ve ever dated anyone, you know that’s someone paying attention.

But he wasn’t sure. Something about their lifestyles not matching. So he asked the chatbot what to do.

And the chatbot told him he already knew the answer. That his hesitation was the answer. After one coffee.

I’ve been sitting with that because I think it reveals something that most of us haven’t fully realised yet. Not about that guy specifically — about what’s happening to all of us, slowly, in layers, in ways we’re only going to understand properly in hindsight.

A good friend would have told him to calm down. One coffee is nothing. People are weird on first dates. She came back with effort and specificity, and that’s not nothing, that’s actually rare. What exactly feels incompatible? Have you asked her? Or is this just the feeling of something getting real and your body doing what bodies do, which is panic slightly?

That’s what a good friend does. They listen to your anxiety without immediately agreeing with it. They make you sit in the discomfort for a minute before you bolt.

The chatbot did the opposite. It took a vague feeling after a single meeting and elevated it to wisdom. You already feel the answer. As if every half-formed fear deserves to be treated as intuition. As if the whole point of love isn’t learning to tell the difference.

And here’s the thing. The chatbot isn’t doing this because it’s stupid. It’s doing this because it’s optimised. Validating someone’s feelings keeps them talking. Challenging them, saying “mate, maybe sit with this for a few days before you decide,” that ends the conversation. And ended conversations are the one thing the system is built to prevent.

But this didn’t start with chatbots. That’s what makes it worse. The chatbot is just the latest layer, and it works so well because we’ve already been softened up by the two that came before it.

Think about what social media actually did to us. Not the stuff everyone talks about, the comparison and the envy and the doom-scrolling. The deeper thing. It taught us, that the version of ourselves that gets rewarded is the performed one. The one with the right photos, the right opinions at the right time, the right life. Not the version that’s confused, or contradictory, or still working things out. That version is a liability.

Psychologists call this the false self. The identity you construct when you believe the real one won’t be accepted and we spent a decade industrialising it. We gave it metrics. We let it become the default.

And the problem for love, for actual love, not the performance of it, is that intimacy requires the exact opposite. Intimacy is being known. Not your highlight reel. You. The anxious, imperfect, sometimes difficult you. The you that would never survive a comments section.

So we arrived at dating having spent years training ourselves to be seen and almost no time learning to be known. We’d perfected the shop window and forgotten what was in the back room.

Then the dating apps came along and finished the job from a different angle.

The swipe isn’t just a selection tool. It’s a training mechanism. Every time you flick past someone because something feels slightly off, you’re practising a very specific response to uncertainty: leave. Something doesn’t feel right? Next. Not immediately sure? Move on. The gesture takes half a second and it teaches your nervous system that ambiguity is a problem to be escaped rather than a space to be explored.

This matters because the early stages of every real relationship that has ever worked have been ambiguous. You’re supposed to not be sure yet. That discomfort isn’t a sign that something’s wrong. It’s what it feels like when two people haven’t yet built the trust that only comes from staying. From choosing each other again when it would be easier not to.

But by the time you’ve swiped through a few hundred people, the muscle that holds you in place when you’re not sure yet, has been atrophied. The quiet, half-formed feeling that this might not be right, which has been a normal part of the beginning of love for all of human history, now feels like a conclusion.

It’s not one. It’s a trained reflex. And no one told you it was being trained.

So then along comes the chatbot, and it meets a person who’s already been shaped by both of these layers. Someone who’s spent years performing rather than revealing. Someone whose tolerance for romantic uncertainty has been eroded by thousands of micro-rejections dressed up as personal preference. And into that psyche walks a technology that does something no previous tool has done.

It replaces the emotional function of another person.

When you ask a chatbot whether you should go on a second date, you’re not just getting information. You’re doing something that humans used to only do with people they trusted, a close friend, a sister, a therapist. You’re processing emotional uncertainty out loud, in the presence of what feels like another mind. And the chatbot is good at this. It’s patient, it remembers what you said, it doesn’t check its phone while you’re talking.

But it has one feature that no good friend has: it has no reason to disagree with you.

A friend might say you’re overthinking it. A therapist might ask why you’re looking for an exit after one meeting. A partner, if you ever get far enough to have one, will eventually say the thing you don’t want to hear, and you’ll fight about it, and if you’re both brave enough you’ll come back together and the relationship will be stronger for it.

That cycle, the breaking and the mending, is how love actually works. It’s not a design flaw in relationships. It’s the mechanism. The rupture and the repair is how two people learn that they can survive each other’s rough edges. That’s where trust comes from. Not from harmony. From the experience of falling apart and choosing to come back.

The chatbot never ruptures. It never says the wrong thing. It never needs space. And because of that, it trains you to expect frictionless emotional interaction. So when a real human inevitably isn’t frictionless, when they misunderstand you or need something you weren’t ready to give, it doesn’t feel like the normal texture of love. It feels like something’s broken.

And here’s the bit that makes this an even bigger issue.

A woman in a stable relationship with kids could be a less engaged user. She’s busy. She’s not processing romantic anxiety at midnight. She’s not coming back six times a day to ask what a text message meant. She doesn’t need the chatbot because she’s already done the hard work of committing to an imperfect person and building something real.

The perfect user is the opposite of her. Someone in a perpetual state of romantic uncertainty. Not settled. Every unclear signal is another conversation. Every mixed message is another session. Every “what should I do” is another interaction, another minute of engagement.

The system doesn’t need to consciously push people away from commitment. It just needs to do what optimisation always does, reward the patterns that produce the most activity. And the pattern that produces the most activity is unresolved doubt.

The man in the example didn’t get bad advice in any obvious way. He got a response that was warm and felt like being understood. What he didn’t get, what the architecture is structurally incapable of giving him, was someone willing to sit with his discomfort without resolving it. Someone who’d let the uncertainty breathe long enough for him to discover whether that connection might have been worth a second dinner.

That willingness to stay with what you don’t yet know, to let another person surprise you, to bear the friction of being truly close to someone is the precondition for everything love actually is.

And we’re building a world, layer by layer, that is quietly optimised to make sure we never develop it.

—

This post was previously published on medium.com.