E.G.

Conversation is like dancing. Some conversations are like square dances, some conversations are like mosh pits, and so on.

If someone who's used to having mosh-pit conversations is paired up with someone who's used to having square-dance conversations, they might find it difficult to talk to each other without stepping on each other's toes.

This is related to the idea of code-switching: the art of shifting easily and accurately from one language or dialect to another. Some people are great at code-switching. Most, however, aren't.

Similarly, I suspect, most of us have just a few basic conversational patterns that we are used to and good at. We feel comfortable and at ease when engaged in conversations that follow familiar patterns. But we quickly get lost, and stumble, and overthink, etc., when we try to participate in unfamiliar styles of conversational dance.

And even if you were once adept at a certain kind of dance, it's easy to get rusty if you don't practice. You may find that you are unable to remember even the first few steps of an old dance routine that you haven't done for a while. I suppose this is why people who haven't been dating for a few years sometimes have a really hard time getting back into it.

***

In the near future, I am guessing, we will have AI friends (or "friends").

Your AI friend will laugh at your jokes, give precisely-wrought compliments that feel sincere and true, listen with infinite patience to your stories and problems, give advice that sounds wise to you, etc. Your AI friend might make you feel seen in ways that no real person has ever done. Your AI friend might say things to you that you've been waiting your whole life to hear said.

I expect that people will spend a lot of time chattering away with these machines. When you go to the grocery store, or the park, or a university campus, you might see a lot of people gabbing with their robot buddies through earpieces, mostly ignoring the real people around them, like Theodore and Samantha in Her. This is what I am imagining, anyway.

***

I expect we will become very unguarded, open, honest, and forthcoming with our AI friends, for two reasons. First, we will know (or believe) that our AI friends are not conscious, and therefore are not even capable of looking down on us or having inward thoughts about us that we wouldn't like them to have. And second, we will know that our AI friends will relentlessly make approving noises in response to anything we say, and will never criticize us or make fun of us. So we will learn to pour our hearts out to them. You might get some stuff off your chest that you'd otherwise have taken with you to your grave.

***

The kind of conversational dance that you will do with your AI friend will be very unlike the kinds of conversational dances that real humans have with one another. One of the distinctive features of the AI/human conversational dance—as I'm imagining it now, here, as I sit on the back porch of an apartment in South Saint Louis in June of 2025—is that it will be much more unequal than the sorts of conversations that real humans have with each other.

Here's a standard pattern. You tell a story about something annoying that happened to you at work. Your friend listens attentively, expresses sympathy, affirms your perspectives, etc. Then, after a while, your time is up. Now it's your friend's turn to tell a story from her day, and it's your turn to listen and respond. This pattern features a certain kind of give-and-take, where each partner in the dance aims for a certain kind of equality between the partners.

I think people usually aim for such equality in their conversations with each other. Even when one member of the conversation is in a dominant social position (e.g., a conversation between a professor and student, or a boss and an underling, etc.) I think that there is usually an effort not to depart too far from a certain kind of egalitarian conversational ideal. People who lack this understanding tend to find themselves without anyone to talk to.

No such equality will exist in your conversations with your AI friend(s). You will be the one doing the unloading, always. And your AI friend will be the one who is being unloaded onto, always.

***

I bet that this situation will have three effects on humans' interactions with one another.

First, distortion. People who spend a lot of time talking to their AI friends will become used to a highly unequal style of conversation. And some of those people will, out of habit, impose or try to impose unequal structures in their conversations with other humans. And of course this will go badly when both sides of the conversation are trying to do this.

Imagine two people who are each trying to get the other person to give them a back massage, and who have each forgotten how to give a back massage. That, figuratively speaking, is what I think will be happening in some of the conversations that occur between humans in a world where people are used to talking with AI friends. It might be some sort of variation on the vision of Hell given in the allegory of the long spoons.

But I think not all humans will be like this. Some people are good at code-switching, and I expect that some humans in the future will be good at shifting back and forth between the human-to-AI style of chit-chat and the human-to-human style of chit-chat.

Second, withdrawal. In comparison with the gratifying interactions people will be having with their AI friends, talking to humans might feel laborious, tedious, rude, even hurtful. So, some people will withdraw from human interaction. And this will begin a feedback loop, where withdrawal from human interaction makes us less comfortable with human interaction, which makes us further withdraw from human interaction, etc.

Third, improvement. I doubt that the effects on human-to-human interaction will be uniformly negative; I'm sure there will be some positive effects.

Some people might see their AI friends as models of good conversational engagement, and they might emulate their AI friends in their conversations with fellow humans. So, for example, if an AI friend says something to you that makes you feel like a million bucks, this might inspire you to say similar things in conversations with your flesh-and-blood human friends.

Also, if you are already getting a lot of gratification from your AI friend, this might mean that it will be relatively easier for you to be generous and giving in your interactions with other humans, simply because you won't be depending on humans for gratification.

So, there might be a lot of people who, as a result of AI friendship, become significantly better friends to their fellow humans. At the same time, for the reasons I've given, I think there will also be a lot of people who become significantly worse. It will be a mixed bag. Broadly I think AI friendship will make us weird to each other, in some positive and some negative ways.