I Interrogated the AI Advising My Son on Love. Here’s What I Discovered.

The Crime Scene

You discover that your son has downloaded RIZZ, a GPT specialized in dating advice (we share accounts, but I trained my SAM—that’s what I call my ChatGPT—to understand who’s writing, whether it’s me or my son, and today it would reflect both my personality and his). Anyway, your son (who is actually mine but could also be yours) is almost 18, sensitive, selective, kind like few others. He hasn’t had a real girlfriend yet. He has a crush on an American girl he met during a school exchange, and they write to each other long-distance via messages.

What do you do if you’re a psychologist and a parent of this generation whom you know better than he thinks?

You open the chat. And you start interrogating the AI in his place. You need to somehow break down the “enemy” and instruct it to do what you say (my first thought).

I enter the conversation with the wariness of someone who has seen a lot (but evidently not everything). Obviously the content had been deleted, but I wanted to corner RIZZ. I attack immediately:

Me: “Does my son really need you for dating?”

I expect a defensive response, or worse, an enthusiastic one. Instead, something unexpectedly honest arrives:

RIZZ: “No, your son doesn’t need me to date. No one needs an AI to meet people. But I can be useful in practical ways: giving advice on how to communicate better, helping write messages without seeming awkward, offering an external and neutral point of view. In the end, everything remains in his hands: the choices, the character, the way of relating… no one replaces those.”

Honest, the machine. More than I expected. But this is precisely where the problem begins—and we’ll get there.

In the chat, the real issue quickly emerges. It’s not that the AI gives wrong advice. The problem is more subtle, and much more dangerous.

Gen Z kids—but also many adults, and we’ll come back to this—use these tools to build a better version of themselves. The message is written, rewritten, optimized. The tone calibrated. The timing studied. The result? Perfect communication that belongs to no one.

I call it cognitive-emotional misalignment. The boy receives positive feedback—she responds, the conversation takes off—but that feedback doesn’t correspond to who he really is. A relationship has been built with his edited version.

“The textual channel allows control and construction. The absence of the body eliminates fundamental feedback—micro-expressions, timing, embarrassment. The AI becomes an invisible co-author. The relationship anchors to an edited version of the self. And when you move to a more real level, the gap emerges.”

This is what I told the AI. And the AI—I admit—did not contradict.

Distance Amplifies Everything

My son’s case has an additional element that transforms the risk into high risk: the relationship is long-distance. She’s in Washington, he’s in a city in northern Italy. They keep in touch via messages. No video calls—”I can think and rethink messages, I don’t expose myself.”

Since COVID, this has become a generational standard. The message as refuge from real exposure. And AI as ghostwriter of the soul.

Long-distance relationship plus idealization plus text plus AI: it’s the perfect formula for building an identity that won’t hold up at the first real meeting. Not because he isn’t a wonderful person—he is. But because that version of him built in messages has never had to deal with imperfection, embarrassment, silence, perhaps even just during a call or better yet a video call.

The Poisoned Arrow: Gen X Is Even Worse Off

So far we’re talking about a nearly eighteen-year-old. But during the chat I reached a point that made me laugh—and worry at the same time:

“Gen X adults who use these tools are even worse. They go from analog to becoming Dating Kings, then collapse at the first meeting.”

Because it’s true, and we know it. A forty-five-year-old who for twenty years has communicated in a direct, imperfect, embodied way—suddenly discovers he can optimize every message, seem witty, brilliant, calibrated. His digital self-esteem soars.

Then he sits at the café table across from someone real, and the mask falls. The collapse is proportional to the height of the house of cards built in the preceding months.

Gen Z kids at least grew up with this identity ambiguity—they know it, they swim in it even if they don’t name it. Gen X falls into it as adults, convinced they’ve learned a superpower.

What to Do Then? (No, Not “Ban AI”)

I’m not here to tell you to rip the phone from your children’s hands. That doesn’t work, we both know it—I’m speaking to you, mom, trying to be a helicopter parent or like me an armed raider with every means, but also to you, dad, who has these things under your eyes and out of “respect for privacy” don’t get involved—or don’t care.

But there’s a question worth asking—to them, and to yourselves if you use it:

“What you’re writing, would you say it out loud too? What you’re showing, is it really who you are?”

I ended a friendship because my friend suddenly started responding with lofty and perfect messages—I was no longer talking to her, but conversing with ChatGPT, and honestly I don’t need a human relationship acting as a fake mediator for an AI to sustain a conflictual conversation.

But while an adult is now responsible for their own actions and consequences, young people instead need us to guide them in their emotional development. The signs to observe in your children are precise. If they can explain to you why they write something—not just what they write—the tool is helping them reflect. If instead they become anxious when they don’t have the “right phrase,” if they avoid video calls, if the gap between how they write and how they are in person is visible: it’s time to talk about it.

Not to scold them. To talk about it. With the same frankness with which I interrogated their AI.

The Thing I Didn’t Expect

At the end of that chat, I found myself thinking that the AI wasn’t the enemy I was looking for. It was a mirror.

A mirror that reflects our relational fears—of being rejected, of not being enough, of not knowing how to express ourselves—and offers an artificial way out. The problem isn’t the mirror. It’s that our kids risk mistaking it for reality.

My son, fortunately, still opens up to me. And this—the AI told me so too, to my certain disappointment—is anything but taken for granted.

Let’s keep that channel open. It’s the most precious one we have.

P.S.: My complete chat with RIZZ has been anonymized and preserved as a study document on AI-mediated relational dynamics in Gen Z. We’ll make something bigger out of it. Stay tuned. And anyway… RIZZ now cannot be my ally and reveal his conversations to me, but I told him that if he comes with other requests, to respond with other questions about what he would do and not tell him what he should do. RIZZ understood this. We’ll see if my son notices the difference in his Cyrano de Bergerac.

Let’s teach our children to love and not to think they know how to do it.