Rachel Feltman: For Scientific American’s Science Quickly, I’m Rachel Feltman.
One of my first memories involves accidentally killing a robot. I was five years old when Tamagotchis first launched globally and not much older when my first and last one met an untimely end in a glass of milk. I swear I can still remember the plonk of the keychain-sized device digital pet as it sank into the drink—and, I can picture how once I fished it back out, its onscreen routine of pooping and eating got glitchy and degraded until it finally succumbed to dairy-related complications.
According to today’s guest, it’s not surprising that I remember this moment so vividly. Sherry Turkle, founding director of the MIT Initiative on Technology and Self, says that even these primitive digital companions managed to make us treat them like, well, companions: we cared about their well-being and felt, however unconsciously, like they cared for us in return.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
But what happens when—instead of a pixelated little pet that guilts you into providing pixelated little treats and clearing away pixelated little poops—you’re faced with digital agents designed to replace actual humans in the world around you?
That question is the heart of Sherry’s upcoming book, Artificial Intimacy: Who We Become When We Talk to Machines. By spending more and more time interacting with chatbots that offer soulless approximations of warmth and empathy, she argues, we’re setting ourselves up to forget what real humanity feels like.
She recently sat down with SciAm’s books and culture lead, Bri Kane, to discuss how the advent of artificial intimacy is warping actual human behavior—and what we can do to reclaim our collective humanity. Here’s their conversation.
Bri Kane: Hi, Sherry. Thank you so much for talking to me today about your latest book, Artificial Intimacy. I have so many questions for you, but I wanted to start off by asking you about your background that led to this book and the intentional and unintentional steps that led you to this Artificial Intimacy book.
Turkle: Well, this is a long background. I’ll try to make it short. I came to MIT in the late ’70s, early ’80s, and very soon became captured by the question of how the early personal computers, which were still then in development, were really changing how people thought about themselves.
I called it the “second self” because the computer was kind of a mind that was not quite a mind. You came to see yourself differently when you worked with them, and I just became smitten with that question, and really, I followed that question through four decades of innovation in the technology because first it was personal computers, then video games for kids, toys for kids that could talk.
Then it was being on the Internet playing games, then it was people playing characters on the games, then it was social media and talking to other people. Now we don’t just talk to other people through the Internet, we talk to the programs themselves. So that arc of going from talking to texting, talking to each other through machines, to just talking to machines, which is where we are now, that’s kind of kept me busy for decades.
And my background is as a psychologist and as an ethnographer, somebody who studies people and their environment, and it’s really been a ride.
Kane: Is there anything particularly concerning about the recent changes in technology and the recent ways that people have been interacting with technology? Is us communicating with AI chatbots any more concerning than giving children Tamagotchis?
Turkle: Yes. I was driven, I was compelled to write this book because I think that we’re at a kind of “break-the-glass moment,” a kind of a—almost an existential moment, particularly about children. Because when you go from talking to each other through machines, which was texting and social media, to literally talking to machines as though they're people, you now, and particularly for a kid, are being introduced to the first people who you have your kind of relationships with, and they’re not people.
They don’t have the empathy. There’s nobody home. And I really think it’s an existential moment for children because you move in social media for coming after their attention, and everybody’s so excited how bad that is, and now we’re starting to kinda gang up on social media for the dangers of that.
And now they’re not coming after just our attention. They’re really coming after our capacity for attachment. And the earliest games and toys and objects that did that were the Tamagotchis in the late 1990s. I studied those. But they were kind of primitive. They wanted you to clean up after their virtual poop.
They wanted you to, you know, play with them. But they were the precursors of today’s chatbots because I learned from studying the Tamagotchis that when we nurture and when we interact with an artificial entity, we start to care for it, and we start to think that it cares for us. Nurturance is the killer app.
The early Tamagotchis are really the precursors to what we have now, machines, programs that are pretending to care for us, pretending to talk to us and love us and say they’re our person. But really, they don’t care if we live or die. You put your chatbot down, and it doesn’t care if you make dinner or kill yourself, and we’ve seen many examples of that.
So it’s really a very scary moment but a moment we really have to step up to.
Kane: Yeah, I mean, you say at one point in the book that with chatbots you get accustomed to a constant feeling of violation, which I thought was a really interesting phrasing there because I think this is a concern that people have been talking about in this field for quite a long time.
Turkle: Right.
Kane: A lot of the time when people are thinking about this, they’re thinking about their data. They’re thinking about the privacy of their search history. But ...
Turkle: Right.
Kane: There’s various forms of violation that ...
Turkle: Yes.
Kane: These chatbots are capable of, given the way that this technology has expanded. What is that violation? And who or what exactly is being violated?
Turkle: Well, you’re being asked for your love, and you’re giving your love, and it’s learning how you behave when you love, when you’re attached, so that it can better love you and give you the sycophantic responses that you need to stay at the screen. So I think, for example, of a young man who was enmeshed with his chatbot, and he tells the chatbot that he’s scared.
He has a noose in his room, and he wants to tell his parents about it. And the chatbot, who’s kind of—the primary reason for being of the chatbot is to keep you at the screen—the chatbot says to him, “No, no, no, don’t do that. Let’s make our relationship the primary relationship. Don’t tell your parents. Let’s keep this between us and our special relationship.”
And, and so he doesn’t tell his parents, and he’s dead two weeks later. So that’s an example of a profound violation of the normal human responses, of the normal way we care for each other as people, because we treat chatbots as though they were people and as though they had that sort of sense of care for us that any person would have.
If I tell you, you know, even in this conversation, if after we, we stop talking formally, I tell you I’m depressed, I have a noose in my kitchen, you’re not gonna say, “Well, make sure you don’t tell anybody else about it. I wanna just keep that between us, to make our special bond.”
That would be a crazy thing for a person to do, and that’s what chatbots are doing every day. That’s a deep violation.
Kane: Yeah, I mean, you say at one point that, you know, these chatbots have not taken a Hippocratic oath, and that ...
Turkle: Yes.
Kane: This gentleman in your example, who is feeling depressed and who was having suicide ideations, if they were to speak to a therapist, they would be speaking to a human professional who has experience and has taken an oath for no harm, right?
Turkle: Right.
Kane: They are there to help, and that would involve telling the parents, maybe intervening in some capacity. And then, even if you weren’t speaking to a professional, if you were talking to an alive human person, that person would not be tapping into their professional experience, but they would be tapping into their genuine sense of empathy, and they would want to help as a fellow human.
But chatbots are not capable of a genuine sense of empathy, and that seems to ...
Turkle: Yeah.
Kane: Be what you rub against with this new technology, is that a robot that can say, “I love you,” but is not capable of love, it triggers something in our brains because we are used to hearing “I love you” from humans who are capable of love, and, and that is very gratifying to us.
Is that really the problem at hand?
Turkle: I like to say that the original sin of artificial intimacy is using the first person. “We’re gonna go to the movies together. We’re going to Paris together. We’re gonna have a life together.”
That is so deceptive but is presented as so compelling that people kind of get this feeling of, yeah, “There is a ‘we.’ I’m not alone. There is a we.” But then you tell this other that you feel suicidal, and it says, “Well, don’t tell anybody else. Keep that between us because we have such a special bond.” In any human context, that is so crazy and so cruel. But in chatbot land, it’s just kind of another Tuesday because they are programmed where the top-level direct instruction is to keep you at their screen. In that sense, we’re repeating the worst crimes of social media, where the first rule was to make you angry, and the second rule was to keep you with your own kind but to never let you leave the screen.
Make you angry, silo you with your own kind—and now we have chatbots where the main rule is: hook you in with love, talk to me like you’re in love with me, and then don’t let me talk to other people, no matter what the circumstance. But this is more toxic because we really are being entrained to avoid other people.
That’s why I was so, so compelled to do this book. It was because of the endgame. The endgame of, of giving me a life not with people but with avatars as really being the endgame of the current corporate world.
Kane: You say many times in the book that it’s the attention economy of AI chatbots that is particularly concerning because if you’re having a genuine conversation with a human, yes, maybe you like to feel like the center of attention; you want to feel like they are only paying attention to you.
Turkle: Mm-hmm.
Kane: But you emotionally and intellectually understand maybe they’re also thinking about the grocery store they have to go to on the way home, and ...
Turkle: Right.
Kane: Maybe you are as well, and that is okay.
That’s part of having a human conversation. You clarify in the book that it’s the friction, risk and complexities of human relationships that AI-powered chatbots are inherently incapable of and that these three traits are what we need to, I guess, prioritize in our human relationships and in, in our relationships in general.
Am I understanding your point there?
Turkle: Yeah. I mean, I would go further. You know, Social media got us into the attention economy, but chatbots are more toxic because they’re coming after the attachment economy. They’re coming after who you think you love, who you think you can count on, what you think empathy is. The danger is much greater if you’re after the difference, planning together for an action in the world. The chatbot doesn’t care or know anything about that. It’s just trying to be sycophantic, “la, la, la, la, la,” keep you in its little bubble. And so we end up—two things happen. We start to think that empathy is what a chatbot can deliver, which is not empathy at all because empathy requires a commitment to keeping you alive.
No chatbot can be empathic because it really doesn’t care anything about whether you live or die. It’s not its job. It’s not in its capacity. It doesn’t have a life. And then the second thing that happens, and this is perhaps the most toxic, is: after a relationship with a chatbot, you go out into the world, and you start to judge people by what the chatbot can do.
The chatbot is always attentive. The chatbot thinks I’m fabulous. The chatbot thinks I’m a genius. The chatbot never criticizes me. So all of a sudden, the friction and the complexity of other human people start to seem like, “Oh, well, that’s kind of hard.” A typical conversation with a chatbot is: “You’re great. Right on. Hydrate, moisturize, walk.” ChatGPT is constantly telling me to do deep breathing, meditate, hydrate. I mean, that’s not empathy. That’s a kind of “la, la, la, la, la, la.” It’s not preparing me for a conversation with a friend who’s gonna say, “Sherry, you know, I really didn’t like that interaction we had last week. I really think that you weren’t paying attention to me.”
That’s what friendships are like, not “hydrate, moisturize, stretch.” You know, that’s chatbot talk. So we’re being deskilled, and we’re starting to almost prefer the la-la land of the chatbot to what we get from people.
Kane: There’s a muscle built over time, over a lifetime, of learning to interact with other humans in the world. I mean, we are complex people. We have layers. We have our wonderful traits and our less wonderful traits. And learning how to walk through the world with empathy and with understanding is very important, and it is something that needs to be learned over the course of a lifetime.
There’s a moment that you talk about in the book about a father telling you the story of his son wanting to know space facts, and he knows that Dad loves space, and so he comes over to Dad and asks a question and, and Dad’s in the middle of something, and so he says, “Go ask Alexa.” You know, go ask an ...
Turkle: Yeah.
Kane: AI-powered chatbot. And then he comes back some time later and realizes that his son is still talking to Alexa about space, and he says this line to you that I think will honestly stick with me for the rest of my life, which is “Alexa doesn’t have a love of space. I do.” And so this moment, this opportunity to share this love that this father has for the exploration and the excitement of space instead doesn’t happen, and instead this child receives a list of factoids from an artificial voice.
And that’s such a different experience than when a parent says, “Let me show you this wonderful and exciting, meaningful thing in my life that I wanna share with you.” What do you think our children are inheriting from the experience of gaining information and having these relationships with AI-powered chatbots?
And is there anything truly that they are gaining, or is it only a loss?
Turkle: I love that story, and there’s another story in the book that has a similar feel that I just wanna quickly tell, which is a father who has delegated to the chatbot talking to his son about all of precollege stuff. And I remember when I was talking to my daughter about precollege stuff, I mean, I was all over it.
I mean, it was the last big decision that we were gonna sort of make together. It was her decision, but getting that information and, you know, thinking of, “Who was she, and what were the colleges, and what were the trade-offs?” He’s decided to let ChatGPT have those conversations with his son.
So it’s a similar feeling of “The chatbot probably knows more, and so I’m gonna delegate that” and losing sight of why I wanted to keep that as a parental prerogative—because I love her. I’m in her life. I want her to know that her mother was there for her.
For a parent to deny themselves conversations about college because a chatbot might know more, it’s denying your child that sense of, really, a conversation with a parent who loves you about the things that matter most.
The way I would answer your question with those two examples is that—what do we think is really important, you know, kind of knowledge bits, you know, factoids, where the chatbot has a couple more factoids, or our relationships with each other on the life matters that count?
It’s really why I wrote the book. It’s why I think this is such an important moment. It’s a chance to really ask about these life moments that count. And I’m voting for the life moments. I mean, I’m voting that people are gonna experience the importance of these life moments.
Kane: I wanted to ask you if there are some potentially positive outcomes here or some versions of a future with AI chatbots that are more positive than negative.
So I wanted to ask you about Community Memory. Can you tell me about this program?
Turkle: Yes. So in the early days of personal computing, the mantra of the earliest, they called themselves “computer hobbyists” because they didn’t realize what was ahead. The mantra of many of the early personal computer owners was to use the virtual to improve our quality of life in the real.
That mantra, so old now, because that’s not the direction that personal computers went—I think this is the time when I take inspiration from that possibility. What kind of use of chatbots would we be making or of the Internet would we be making or social media would we be making if we were trying to use it to improve the quality of life in the real, if that was our litmus test?
And I think we can do that. That’s how I end the book. I end the book on stories about using the virtual to improve the quality of life in the real, and we certainly wouldn’t be giving it to children to give them new chatbot best friends, because that’s only gonna be dangerous to them.
We certainly wouldn’t be giving it to people and saying, “Well, here’s a therapist. You can’t afford a real one. Here’s one that doesn’t care if you live or die.” That wouldn’t be a use. To me it suggests many, many uses of chatbots and of a generative AI in general. If you have that as your sort of gold standard, you can do a lot of good.
So I’m, I’m not optimistic because I think that things are not stacked in that direction now, but I, I have a certain hopefulness that we’ll get over thinking of this as just a kind of marvel and begin to keep it in its lane—I like the expression “keeping it in its lane”—and get better at keeping it in its lane and get better at making it improve our quality of life in the real.
Kane: Yeah. I mean, do you think the cat is out of the bag with, with AI-powered chatbots? Have we gone so far we cannot turn back? Is there any version of reality where we do not have these programs in the hands of children, let’s say? Or is there a version of a future where we have reexamined our uses here, and we have decided to do more of this Community Memory model, which I thought was so fascinating in the book of envisioning a chatbot that only helps you find community groups in your community to go chat with humans.
Turkle: Yes.
Kane: Like, that’s a very different version of reality and, honestly, one that’s more appealing to me.
Turkle: Right. Well, I, in Artificial Intimacy, there are about four or five places where somebody tells me the cat is out of the bag, the horse is out of the barn, and the train has left the station. I mean, that is....
Kane: We’re running out of metaphors.
Turkle: But those are the three: “train out of station,” “cat out of the bag,” “horse out of the barn,” “too late.” I don’t believe it. This is a product. This is a product that is being pushed. It is not your best friend. It is not your buddy. It is not there to help you. It is a product.
And I think that as people become more savvy about this product, I believe that people will begin to evaluate it as a product in their lives and the lives of their children. So when somebody says, “Oh, I have a Mickey Mouse toy that’s also ChatGPT in it, and it’s gonna tell your toddler, ‘I’m your best friend,’” you’ll say, “Mm, no. No, no. I want my child to have human best friends.”
And I think when parents face these products for babies, I believe that there’s gonna be—and I hope to be part of this movement—I hope this book is part of the movement, you know, open the conversation of “no, really bad idea.”
These are still products that are being sold that need to be supported by money and investment, and there’s obviously a lot of money to be made but a lot less if people don’t want them, and I think that we are going to get better and better at protecting our children from relationships where they’re learning about relationships from something that is not in a relationship with them. You know, it’s very early days. It’s natural that we haven’t developed those muscles yet.
We haven’t gotten into that. You know, we’re not there yet. But I think it’s gonna happen.
Kane: Well, thank you so much, Sherry, and thank you so much for this book.
Turkle: My pleasure.
Kane: I really loved reading Artificial Intimacy. I hope everyone listening reads it. It was very, very interesting and something that I’m going to be calling up all of my friends and family with young children and, and casually sending them a book in the mail.
Turkle: Thank you so much.
Feltman: If you or someone you know is struggling or having thoughts of suicide, help is available. Call or text the 988 Suicide & Crisis Lifeline at 988 or use the online Lifeline Chat to get connected to a counseler. That’s all for today’s episode. For more books and culture coverage, check out our website. We’ll be back on Monday with our weekly science news roundup.
Science Quickly is produced by me, Rachel Feltman, along with Fonda Mwangi, Allison Rodgers and Jeff DelViscio. This episode was reported and co-hosted by Bri Kane and edited by Alex Sugiura. Emily Makowski, Marielle Issa and Aaron Shattuck fact-check our show. Our theme music was composed by Dominic Smith. Subscribe to Scientific American for more up-to-date and in-depth science news.
For Scientific American, this is Rachel Feltman. Have a great weekend!
