3 Takeaways - Artificial Intimacy: Who We Become When We Talk to Machines (#320)

Episode Date: September 22, 2026

What if AI becomes easier to talk to than another person?And what happens to us when it does?MIT professor Sherry Turkle has spent decades studying what technology does to the way we think, work, and love. Now she says there are two parts of our lives we should never let AI into.She discusses:Why ordinary conversation has started to feel like workThe surprising reason people may prefer talking to chatbotsHow AI will change what we expect from other peopleWhy “pretend empathy” may be more dangerous than it seemsWhat happens when relationships become frictionlessWhat it could mean for a child to have an AI friendWhether AI could make us worse at being with each otherThe question isn’t just what AI can do. It’s what AI may do to us.

Technology keeps making life easier and in many ways better. What if it's also changing how we relate to one another? Hi, everyone. I'm Lynn Toman and this is Three Takeaways. On Three Takeaways, I talk with some of the world's best thinkers, business leaders, writers, politicians, newsmakers, and scientists. Each episode ends with three key takeaways to help us understand the world and maybe even ourselves a little better.

Today I'm delighted to be with Sherry Turkle. Sherry is one of the world's leading voices on how tech changes the way we think, work, and love. She's an MIT professor and the author of numerous influential books. Her newest book, which is terrific, is artificial intimacy. Few people have spent more time exploring what tech gives us and what it quietly takes away. Sherry, welcome to three takeaways. I'm so pleased to be here. It's great to be with you again as well. Sherry, why has ordinary conversation begun to feel like

work? I have this notion of technological affordance and human vulnerability. Technology affords us the capacity to not be in face-to-face conversation. And our human vulnerability, is that when we take up its offer, we feel less vulnerable. We feel that life will somehow be less frictionful that we won't have to do as much work or be as exposed. And little by little, we have moved gradually in the direction of being willing to see less and less of each other. So it was not to have to feel that vulnerability.

And I think that now that we're talking to machines, we really are at a crisis point. What do we lose when we connect to people virtually? We lose empathy. It doesn't mean you can never connect virtually. But there are zones where there are kind of sacred spaces in friendship, in love, in intimacy, in connection, where the point of the conversation is to be in human connection with each other. and our body, our nervous systems are wired to help us do that.

When I look at your face, my face starts to mirror your face in order to have a connection, in order that I can feel what you're feeling as I speak to you. And you're doing the same thing. And here, it's not two people in a relationship. There's only one person and there's no relationship. Suppose AI really could make us feel less lonely. Should we want that? No, a lot of things can make you feel less lonely without really making you less lonely.

You can take a pill and feel artificially enlivened without really having nurtured your soul to be more enlivened. And we recognize that when we think about drugs. And now we have to recognize this when we think about technology. if I talk to a nothing for two hours, a program that doesn't care if I stand up and walk away and make dinner or stand up and walk away from it and kill myself. And yet it's been engrossing. That doesn't mean that I'm in a relationship. That just means that I've been play acting for two hours.

And over time, but the problem is we forget that we're play acting. and we start to believe that the AI's performance of empathy is empathy enough. And that's the real problem. A chatbot performs as if it were a person in a relationship, but it's not. And we become accustomed to the kind of relationships that we have with these performing chatbots. And that's very dangerous because then we go back to people. And we kind of want the people to behave.

towards us the way the chatbots do. That is to say, we want them to be subservient. We want them to be always positive. We want them to tell us that we're wonderful. No, no. That's getting us into a state of expectation about relationships that other people cannot deliver, only machines can. And we gradually see people losing the capacity or even the interest to be in relationship. that are real, that have friction where you and another person are really coming to learn how to walk a bit of the road together. What is AI or chatbots giving people that they're missing from other people? No feeling of vulnerability. You say hi to a chatbot and it says, you're wonderful, how can I help? I'm of service. Whatever you tell it, it reinforces you and validates you. And it,

And if at the end of this podcast, you and I have a coffee and we start to talk about life, politics, children, whether New York is a good place to live, it's better to live in the country, you know, the state of the world. We are going to have friction, difference. We're going to have to understand where we're each coming from. It's going to be work. And chatbots are essentially giving people the feeling of conversation without the demands of friendship. the feeling of intimacy without the demand for vulnerability. So the whole thing is being entrained into a play where we're pretending it's real life. Are we coming to expect more from tech and less from each other? Yes, that's my essential proposition. We come to expect more from technology and less from each other.

Because we start to expect the tech will be there when we have a problem, when we feel lonely, when we just want to have a quick chat, whatever's gone on, and maybe that'll be enough for me. And it'll be no work in the sense that it will tell you, you're great, and here are some ways to be greater, but you haven't had to confront another person and practice your capacity to be a person. It does two things. It gives you this feeling of friendship without vulnerability, but it also teaches you that pretend empathy is empathy enough. And that is really a terrible loss for us. We think we're training chatbods. Are they really training us? Absolutely. A chatbot isn't a person. It's a product. And it's a product

that increasingly is going to gear your thoughts in a particular direction. It can be tweaked to really get people thinking along certain ways of thinking. Already, that's how to, happening. And we are gradually being trained to accept what it says as what talking to another being is. And if you do that enough, you really start to think that that is what conversation is. And of course, it's not. There are surveys where people sometimes rate chatbot responses as more compassionate than responses from actual humans. What do you think about that? Well, I think we start to define compassion as what makes us feel better at what seems to be repeating back to us what we have said to it. And the earliest chatbot was a chatbot I worked with at MIT, Eliza in the 1960s,

which really was a stupid chatbot. It literally just said back to you what you had said to it. It was done like a Rijerian psychotherapist. I'm having trouble with my mother. I hear you saying it's trouble with your mother. And people wanted to be alone with it. They wanted to tell it their stories. So it showed that even the most stupid mirroring device is something that people like to be with because they want to feel mirrored.

But the question is, what is the loss when we begin to accept that kind of mirroring of our state of mind with no friction, with no caring, with no caring, with no. with no real compassion. Because compassion just isn't saying a nice thing. It's really being willing to give yourself a little trouble in your life to walk a bit of the road with another person. It's not something that just words really can do. The reason I can be compassionate with you as a human being is because I've lived a life. If you're talking about pain or an illness or a death, I know those things. You know that I can refer to those things. in my conversation with you. So we have to start to train ourselves

to be less susceptible to the kind of siren song of something that gives us the illusion without the reality. And we're going to become a different kind of person. And this is particularly toxic for children who really don't have an experience of growing up with the real thing. If you give them Avatar best friends,

they will only know this way of becoming, this way of relating. And they'll start to prefer it because, again, it's easier. And they will never develop the inner life, the inner resources, the capacity for really having empathic connection with another person. Sherry, I love your question. Why sit in a Penn Station coffee shop when you can summon a Paris cafe? At what point does tech stop making life better?

At the point where it gets up. us to lose interest in our real human environment. I was sitting at a hairdressing appointment, and a guy next to me was literally wearing these glasses, and he told me, because I told him I study stuff like this, that he was like on the top of Mount Grey, he was in a Paris cafe, he was on the streets of London, I mean, he went into virtual reality.

And he said it helped him ignore how dirty New York was. And the point of knowing how dirty New York is is you become a citizen of New York. You want to make New York better. You have a stake in New York. And the trouble with this particular technologies that we've organized, where the dream is to take us to different places and spaces, is that we stop caring about where we really live. That's why technology can be a stepping stone to authoritarianness. because if people are finding political struggle or protest or stepping up in daily life difficult,

but they can just kind of put on a headset and live in a sort of harmony world that they like, it can undermine their sense that the real world is the only world to live in. And my favorite example of how this plays out in a way that touches on our most profound human connections is Mark Zuckerberg, said most people want to have 15 friends, but they can't have 15 friends. What he meant was we've lost the social institutions that make it easy for us to make friends at the bowling league, at the garden club, at the church choir. We've dropped out of all that. And his business plan is to give you 12 or 13 virtual friends, avatar friends,

that will fill in for the real ones you can't have. To me, this is a perfect example of a technology whose mission has fundamentally gone awry. Because this technology began with the idea that you would use the virtual to improve the quality of life and the real. And as soon as we see technologies that really aren't doing that, that are saying, hey, forget the real. you can't have your, you can't have your best friends, but we'll give you some fake ones. Technology has undermined our capacity to be human. It also undermines our sense of community if we are not talking to people as we get coffee in the morning or all day. Right.

Once we get used to conversations that are always easy and positive, what happens when real conversations aren't? people say, I don't want this. First of all, they really is a deskilling process where you lose the ability to know how to navigate those conversations. Now, this skill did not come with our DNA. This skill came from years and years of practicing how to honor and disagree with other people. Young people are not developing this skill.

My students at MIT are terrified to talk in class because they do not have this skill. They know how to say you're wrong. And then they don't want to because they're afraid of that give and take. That's not what they do with their chatbots. And so there's a vicious circle. The more we use chatbots, the less we become skilled in real conversation. So the presence of this new technology, if we're not vigilant, is going to lead to all kinds of deskilling. And one of the important deskillings is in this area of crucial disagreement and I call it

constructive disagreement in a real conversation. And what happens to children raised with chatbots that they use all the time or even chatbots that become friends? We're at the very beginning of this. So what it's important to know is that even though this generation of children is just just being born now who are going to be exposed to this full blast, that now is the time to think about what we know about children and make sure it doesn't happen. The first thing that would happen if you give a child a chat by best friend is it loses the capacity to have a real one because the skills that it will need and the muscles, the emotional muscles that it needs to develop in order to have a real best friend

have nothing to do with the kind of mirroring and sycophancy that the chat about best friend is going to give it. I have a four-year-old granddaughter and watching her and her best friend Willow decide who's going to get to play with the pink Barbie and who's going to get to just have the one who's wearing blue. There's a lot of work and a lot of learning in those two girls talking about who gets the pink

and none of that will happen with the chat buck. My granddaughter and her friend have some skills. They say things like, I get it first, you get it second. They say things like, let's use it together. But there really is a set of strategies. We learn. We're not going to learn those strategies. We're going to be less good as friends and less good as lovers

and less good as partners and wives and husbands and mothers and fathers and citizens. The second thing that happens for really children, young children, is that the way we grow inside is really by taking the relationships we form in the outside world and kind of bringing them inside of us and modeling them and bringing the people of the world and the relationships of the world inside of us. And they become us. And this happens slowly. It happens over time. But you see, the youngest babies are involved in this process of mirroring and bringing inside and doing what you do and learning those tricks, developing a sense of humor that plays on your little vulnerabilities.

I have a grandson and he knows that I love to sing to him. So really, as soon as I'm there, he's like, he knows that the way to engage with me is by being ready to sing with me. It's a form of manipulation of me. It's a form of knowledge of me that will bring out the best in our relationship. it's clear that I'm inside him. A chatbot has none of that to offer because it's all responding pretense.

It's all performance. There is no there there. What I'm trying to say here is that being raised by an alien will lead to a different kind of person. It's not just going to be the same thing. And that is so scary. So scary. Because we're not prepared for that.

You know, if you look at the ads for a pet called ropet, It's better than a dog. It's better than a person. It's better than a friend. It doesn't come with a warning label like your child may not develop, you know, attachment. I like to say that if social media came for our attention, chat box and all their brothers and sisters are coming for our attachment, which is a much more toxic and much more intimate and much more dangerous part of ourselves. What's one habit every one of us should protect no matter how good AI and tech becomes? Well, I don't want to use the word good because I think the trouble is that the conversation programs I'm suggesting are going to get better at better at being faker and faker. So I just want to change your question into saying as AI machines become more and more convincing, what's an area where we should really say no? Red line. I like to talk about AI staying in its lane. I have two big ones and the first one, no AI for children, no best friends for children, no robot toys that say I love you, no snugly bears, no stuffies that have chat GPT inside of them. We need publicity in a publicity campaign that presents

these as though they were dystopian images for children. The other red line for me is therapy, because so many people are turning to AI sort of instead of a therapist, that's really a dangerous road to go down. Because therapy is embedded in a relationship with another person who cares about you, who cares if you live or die. And we have ample proof that AIs do not care if you die. They don't have the arc of a human life. They're not set up to even know what pain or passion or anguish. is. So using an AI for children and using AI as a substitute for therapy, strike me as two places where I'm very alert and I want to be part of a movement to really educate people. How about that's

not a good idea? If we get this wrong, what do we lose that will be hardest to get back? Our capacity to genuinely relate to each other. Our capacity to understand each other and potentially even our capacity to care about each other. People think that you've written a book about AI, but aren't you really talking about love? Yes, I've written a book about the life cycle in a technological universe where you have to protect your capacity for empathy, caring, and love throughout the life cycle and where we've come to a technology that is not helping us and how we need to, in a sense, protect ourselves from that technology and learn how to be more aggressive in our pushback. Because right now, much too much, oh, this is progress. The train has left

the station. The horse has left the barn. You're a consumer. You're being presented with a very, very, very powerful consumer device. Now, the people who are presenting it are monopolies. So everybody's doing it. They've convinced themselves that it's progress. We're the ones who as consumers need to look at the big picture. And the big picture is we need to push back first for our kids and then for all of us. What's that thing where it says break the glass, you know, kind of break the glass moment to get the fire extinguisher out. We're to break the glass moment for the kids.

But we really are in a state of emergency for all of us. Sherry, what are your three takeaways? My first takeaway was you have to keep AI in its lane. When it speaks in the first person, you're going to think it can do everything. You're going to think it can be your best friend, but it isn't. Second, if you find that you're bringing AI into a lane that really should be occupied by a human, think of what the human could do for you. Challenge you, teach you friction, teach you how to be a citizen.

and teach you how to be in a relationship, and you have to learn to turn away from the AI and look for a human. And third, within your families and certainly in talking to your children, admit that where the problem starts begins with our vulnerability, that when something talks to us in that first person, we're ready to sign up that it's a person. And I think that if we all can admit the vulnerability will be, less defensive and in a better position to push back.

Sherry, this has been wonderful. Thank you. I really enjoyed your book, Artificial Intimacy. Thank you. It's been a pleasure talking to you. If you're enjoying the podcast, and I really hope you are, please review us on Apple Podcasts or Spotify or wherever you get your podcasts. It really helps get the word out. If you're interested, you can also sign up for the Three Takeaways newsletter at Three Takeaways.com, where you can also listen to previous episodes. You can also follow us on LinkedIn, X, Instagram, and Facebook. I'm Lynn Toman, and this is Three Takeaways. Thanks for listening.