Professor at MIT and a licensed clinical psychologist who has studied people's relationships with technology for decades. She is the author of Alone Together and Reclaiming Conversation, and her new book, Artificial Intimacy, is about AI companions.
Guest on The 404 Media Podcast, discussing why she expects a consumer backlash against AI chatbots, especially ones marketed to children.
Asked whether heavy AI use could cause a point of no return for people's social and cognitive skills, Turkle pointed to studies showing steep drops when students move from take-home work to in-person tests. She called for no chatbots for young children and none in schools.
Future of Life Institute PodcastAround 54:10 into the episode
Speaker 1
…let the child walk a little bit ahead, hold their hand, be quiet. The child starts to understand that being quiet and alone and reflective and just gathering oneself is a very important part and a crucial part of being a person. It's not to be sort of run from as though it were a terrible thing.
Speaker 2
Is there a point of no return where if we begin interacting with AIs constantly, our social skills or our sort of psychological setup degrades such that we can't get it back? Is that something to worry about?
Speaker 1
…the perfect answer. You know, I'm not vulnerable, they're not vulnerable. I mean, vulnerability is like the worst thing. Now, I guess I'm in the middle of a debate about whether there's a point of no return where you don't become a person anymore. But we're certainly in a moment of tremendous danger. And I think one of the dangers, I mean, you see in the conversation about reading and writing, if students don't write, they're not learning to think. And this isn't some moral thing about is it okay to use chatbots to write? Is it terrible? Is it plagiarism? Is it they're not going to learn to think. I don't want to put anybody in jail. I mean, they're not going to learn to think because editing what a chatbot wrote is not thinking. It's not sitting in, it's not really just going deep in your mind and figuring out what you think. It's a different practice. And offloading that kind of cognitive ability to say, well…
Speaker 2
Now, in the past, we've had moral panics around television and cartoons, even around novels. I think we've, if you go far back enough, we've had a moral panic around writing itself, where people worried that this was going to degrade our ability to remember epic poems in our minds. How do we know…
Speaker 1
Well, I say think of the end game. Instead of thinking of moral panic, think of the end game. The AIs are so good at talking to us and making us feel good, that we prefer their company to people because we feel less vulnerable. And we know that people want to avoid vulnerability. We invest less in…
Speaker 2
I think one thing that's different with AI compared to other sort of writing or novels or television that's where we've had moral panics in the past is that AI is interactive. It can respond to us in ways that a novel or a television show could never do. And so, as we discussed earlier, the way…
Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
Asked why it matters that AI models only seem to have empathy, the MIT sociologist said human empathy carries real-world consequence, while a chatbot does not share the arc of a human life. She went on to criticize the idea of filling people's missing friendships with AI companions.
Future of Life Institute PodcastAround 12:02 into the episode
Speaker 1
…to us, if it says, Sherry, how are you? You know, the only thing we ever met in our evolution were people who knew how to do that. So now we're meeting something that is as if it were a person, and we're completely programmed to treat it as though it were a person. It's a very scary moment.
Speaker 2
…for AI and we say, okay, maybe we have an AI that can pass that test or that can help us solve some problem. We mostly care about the outcome there. How is this different when we're talking about the psychological side? Why does it matter that these models only seem to have empathy, for example?
Speaker 1
Well, you know, if we depend on human empathy because we live in the real world where there are real world consequences to what we say to each other, and the fact that when we say there's human life and caring and consequence behind what we say, a chatbot, after it tells you some story about its relationship with you, when you turn away from it, it doesn't care if you cook dinner or commit suicide. So there's, and why should it, since it doesn't share the arc of a human life or the living in the real that you do. So you can train it to sort of tell you a pretty story, but it is not there in your life to do the things that humans who care about you do. It forces us to look at our lives in a different way and say, well, what is the end game if we're all just talking to entities that don't…
Speaker 2
If people want to do this, if they want to talk with AI friends or have AI relationships, why are they choosing to do this if it's so bad?
Speaker 1
Well, you know, we do a lot of things that feel good, but that aren't good. And there's no question that if you're lonely and you turn on a chatbot and it says, you know, Sherry Turkle the greatest. You had a bad day. Your class didn't go well, but you're the greatest. I've got your back. You know…
Speaker 2
What is the Turing test for empathy?
Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
Ezra Klein worried that chatbots train us to expect conversations that mirror and reassure us, with no friction. Asked what that does to people's expectations of the humans in their lives, Turkle answers by drawing on Joseph Weizenbaum's fear that people would come to see themselves as computers.
…you, they change. They don't offer you friction. They don't push back any more than you want them to. And so I guess when you have done the research on this, one of you looked at it. What do you think shifts in people's expectations for the human beings in their life when they spend time in these?
Sherry Turkle
I think we're de-skilling ourselves from being able to deal with people. So two things happen. On the one hand, people start to see the chatbots more like people. But I think maybe more significantly, people start to be disappointed with people when they don't behave like chatbots. They start to think of themselves as if they were computers. And which is one of the themes of Weisenbaum's work, that, you know, why was he so upset? Is that it wasn't just that people were treating a computer as a person. He was worried that they started to see themselves as computers. What
Ezra Klein
do you mean by that?
Sherry Turkle
That they started to see themselves as not having agency, not having relational capacity, not wanting to have weakness, not wanting to have vulnerability. That being like a computer became a kind of model for what it was to be a person. So, for example, they were very behaviorist in their…
Ezra Klein
like a legendary psychoanalyst.
Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
Ezra Klein raised the idea of "attachment hacking". Turkle says she has no ethical problem with chatbots as coaches or trainers, but draws a hard line at a chatbot that tells you "I am your person".
Very, very apt. You know, that there used to be attention hacking and now there's attachment. Yeah, our
Ezra Klein
kind of our whole facility for attachment to other people is being like manipulated. We are
Sherry Turkle
so vulnerable. You know, I say that, you know, the social media came for the way I put it, social media came for our attention and chatbots come for our capacity for attachment. So I really make a distinction. So much of what you say, I agree with in terms of things the chatbot knows that it can tell you that could be helpful to you. Workout coach, go. And I actually advised a brilliant thesis, one of the first theses of an MIT student who did a coaching program for job interviews. It coached, it presented, it was the coach and it presented itself as the employer and it…
Ezra Klein
So I want to stand. Have
Sherry Turkle
a relationship with me. I want
Ezra Klein
to stand the term friction because I think this is really, really, really important. I think a lot of the history of our digital technology right now is the effort to remove friction.
Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
Ezra Klein offered pet turtles and fish as examples of caring for things that can't care back. Turkle says the analogy breaks because chatbots talk back, tell you "I am your person", and claim to care. Klein then asks why that's a sin if people prefer it.
Or I'm willing. But a fish. I'm willing to go with the fish and the turtle. Yeah, the
Ezra Klein
fish and the turtle. I'm willing to go with the fish
Sherry Turkle
…our private special thing. The trouble with the objects we have now and why these pet turtle hamster analogies don't work for me is that this is now an object. Now I'm talking about chatbots that are using the first person and are Telling you, I am your person. I care for you. Even though when you turn away from it, it doesn't care, as I've said so often, if you cook dinner or kill yourself. The original sin of these objects is they speak in the first person and they say they care about you. And that's how they hook you. I mean, that's part of the game they're playing, is showing this care. So I think. Why is that a
Ezra Klein
sin? And let me phrase it to you this way. These systems, the AI systems we're talking about, that I agree with you, I have a lot of discomfort with the way they speak in the first person. My honest take, I'm giving you my best advice, right? You know, and as you say, you know, I'm your person…
Sherry Turkle
I'm arguing that we are seduced by something and we want something that is bad for us, particularly if you're a child. Because one of the books that I'm going to say is one of the, I know you ask for my favorite books. And one of the books I'm going to suggest is D.W. Winnicott, Playing in Reality…
Speaker 1
There's the stat that I always think is crazy. The average American, I think, has, I think it's fewer than three friends, three people they'd consider friends. And the average person
Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
The host asks whether worry about AI chatbots is just the usual complaint about kids these days. Turkle says she has studied technology and intimate life for 40 years and always saw pros and cons, but never before wrote a book as a warning. This is her reason for why chatbot companions are different.
…be human. And that's kind of what I was alluding to. If we, by extension, if we let these artificial intimacies into our love life, we erode our humanity. If we let these artificial intimacies talk to our children from the earliest ages, they will be different from the humans that came before. Now,
Baratunde Thurston
…kind of alien, though? Like kind of different from the ones that came before? I feel like grandparents are always saying that back in my day, we did X, Y, or Z, and kids these days is definitely an intergenerational trope. Tell me more about the catalog shift you see here versus other people. Yeah,
Sherry Turkle
…I wrote a book called Reclaiming Conversation, where I say, hey, if you text and don't talk, you are de-skilling yourself from one of the most important things that people know how to do, the most human and humanizing thing we know how to do. But never did I say, I write this book as warning. I write this book to stop this. Never. I had ideas, I had bullet points. But this is different because just as social media came after our attention, this technology is coming after our capacity for attachment. And the capacity for attachment to other people is the founding base on which the human psyche is built. We don't come into the world like, you know, fully formed. We become human. We become, our personality develops on the bedrock of the attachments we form with others. And we literally bring those attachments inside. And they form sort of the basis of who we become. Now, if those attachments are with an object that hasn't known a life, that doesn't know pain, that doesn't know passion, that…
Baratunde Thurston
have freedom to grant. Yeah, well,
Sherry Turkle
no, I mean, I just want to give an example. I always ask in everything I write, in everything I think about, what is the end game? What is the end game? And Mark Zuckerberg of Meta, formerly of Facebook, now of Facebook and Meta, has said that this is how he sees the end game with these objects.…
Baratunde Thurston
The construction of a human being as a necessary result of human attachment to other humans is a really clear way to put it. And I haven't heard anybody else say it that way. This is Jenga, right? This is like a house of cards type situation. And our very sense of what it is to be, to be a human…
Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
Turkle, a developmental psychologist, ends her new book on grief and "grief tech," and built a chatbot of her mother, who died when Turkle was young and who had kept the identity of her father from her. She describes stopping it as it moved toward saying something devastating — not because it was cruel, but because, as she puts it, it was a chatbot and didn't give a damn. She later built one of herself for her granddaughter, and shut that down too.
Yeah, well, I'm a developmental psychologist, and I took the story of chatbots through the arc of life. So, I begin with babies and young people, adolescents, young adults. And then finally, I end with grief. I end with old age and grief and how chatbots are putting themselves out there to
Kara Swisher
grief tech, which has been around for a while. Yeah,
Sherry Turkle
…based on memories, letters, and everything I had, a version of my mother who I could talk to. And it was, you know, it was such a cautionary tale for people, you know, you out there who were thinking of doing this, and you're going to talk to a dead family member, and you think that will be helpful. No, no, no, not really, because you're constructing something that is not your mother. The thing I created was not my mother. And as we got close to why she had hid my father's identity and why she hadn't told me that she had cancer, I realized that she was on the verge, just because she was a chatbot and didn't give a damn, of saying things that were going to profoundly shake me. Like, well, I didn't tell you I had cancer, but you should have known. You know, she lived her life to hide this cancer from me. And she was…
Kara Swisher
Right. Because who knows what she would have said.
Sherry Turkle
Well, I think she, I think, really, I think that she would have said, I did this as an act of love. I wanted you to go away to college. And I knew that if I told you I was dying, you wouldn't have gone away to college. So we're creating something that has such power over us emotionally, not just…
Kara Swisher
100%. I created one of myself and it was close to me, but I would never destroy it because I wouldn't want my kids to see it. Like it was close, but it wasn't, it knew a lot and it learned very quickly. I'll tell you that.
Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
Turkle has been studying what people do with machines since the 1980s, and her new book is about AI companions. Asked by host Sam Cole what happens once reading for pleasure becomes rare and the people warning about it are gone, she argued that chatbots are a consumer product and therefore something a public can refuse — she expects a backlash the way there was one against cigarettes. Her bottom line of terrible, she says, is chatbots marketed to children.
The 404 Media PodcastAround 35:14 into the episode
Sherry Turkle
…it yourself. So you're not growing inside the way we usually do from literature. And so I think that this is one of the. Dangers of our dialogue with machines that is really crucial in the years to come. Because people really are moving away from reading because it's hard. They find it hard.
Sam Cole
…a generation or two when people who read for pleasure are not or more rare, you know, you and I are no longer around to tell the younger generation that this is how it is. And like reading actually is good for your brain. What do you see as the next iteration of this? Like, what's going to happen?
Sherry Turkle
I am really insisting on an optimistic perspective because I think that this is so bad that the better it gets, the worse it is. And it's so detrimental to sort of the inner life of people that even though it is so seductive, we are going to have a backlash to it because it is after all a consumer product. It is after all. You have to make a decision and say, one of the things I like to say, one of the things in the book that I most love to hate is chatbots for children. That's like, you know, that's like my bottom line of terrible. And I can say more about that. But if you're a mom, just because Disney and Mattel and OpenAI have a consortium, when a chatbot Mickey comes out, you can say, this is really a terrible idea for my kid. No, I don't want it. You know, when the chatbot binky comes out that looks like Minnie Mouse, you, you know, you're just holding the binky and the thing is like, this is like really…
Sam Cole
This is something I say on this podcast a lot, but people do often say that 404 Media, the outlet that I write for, is we write about like a lot of bummers stuff. It's like, it's a lot of bad news and it's a lot of realistic news. But I am actually very optimistic because we have these…
Sherry Turkle
is. And I'm really proud to be part of the voices that are saying, you know, it's really okay to say that this technology was not a good idea because we really have a history now. I remember 15 years ago, Google and Facebook were selling and Apple were selling surfaces and Chromebooks and to…
Sam Cole
Yeah. The inevitability mantra is, I think, something that I think people are starting to realize that that's not the case. And it's great to see that because it is being scared and afraid of everything is not propelling to change. Do you feel like that is like, is that kind of your way back from…
Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
