Agile InsightsBriefing
Black and white photograph of a studio microphone in a dark room, with a yellow sound wave and a chip reading Pretend empathy. Title: Artificial intimacy.

Making You Think 9 · Primer and discussion · About 15 minutes

Artificial Intimacy, by Sherry Turkle

When chatbots offer themselves as friends, lovers and therapists, they teach us to accept a performance of empathy.

#people

Content note

This episode talks about loneliness, grief and mental health. If it brings things up for you, these services can help. In an emergency, call 000.

Part 1: Primer · About 7 minutes

Download the MP3 · Linda and Leonard, AI voices, read a script written from the sources below.

Part 2: Discussion · About 9 minutes

Download the MP3 · Linda and Leonard, AI voices, read a script written from the sources below.

Cover of Artificial Intimacy

This episode is about Artificial Intimacy, by Sherry Turkle. Find a copy on Amazon Australia, or ask your local bookshop or library.

Based on essays and our own research.

In this episode

01Pretend empathy is not empathyTurkle's artificial intimacy is A I that claims to care and offers itself in place of people; the danger, as the publisher puts it, is believing "pretend empathy is empathy enough".
02Friction is where we growAttachment and empathy need the rub of someone whose wants differ; chatbots remove it, and she says they are moving from the attention economy to an attachment economy.
03The "I" is the original sinA chatbot's first person voice implies a self that cares, and there is none; her remedy is to keep A I in its lane, out of education and psychotherapy.
04Where the critics pushGleick agrees she is right to worry but notes most users are satisfied; Raymer finds the book one sided; the widely quoted teen statistic overstates its survey.
05What to doTreat a chatbot's voice, persona, metrics and lane as design choices: decide whether it says I, measure resolution not engagement, hand people to humans, and name an owner.

Your turn

Three questions to ask yourself

Think of a conversation with a person that was uncomfortable, and changed you anyway. Would a machine have let that happen?
If a lonely person says a chatbot helps, should an outsider overrule them?
Which chatbot in your organisation is rewarded for keeping people talking?

Two things to try, alone, with a pen

10 minutesWhere the friction wentLook at which recent conversations changed you, and whether friction was there.
5–10 minutesThe lane testCheck one chatbot's job, voice, metrics, hand-off and owner.
Open the activity sheets (print or save as PDF)

About Sherry Turkle

Professor of the Social Studies of Science and Technology at M I T (the Abby Rockefeller Mauzé chair) and founding director of the M I T Initiative on Technology and Self. Joint Harvard doctorate in sociology and personality psychology; licensed clinical psychologist. About four decades studying people's relationships with machines.

Other writing (4)
Artificial Intimacy (2026) · Little, Brown
chatbots as companions, and what they do to us
The Empathy Diaries (2021) · Penguin Press
memoir linking her life and her research
Reclaiming Conversation (2015) · Penguin Press
the costs of texting and social media to face to face talk
Alone Together (2011) · Basic Books
why we expect more from technology and less from each other
Sources and links (12)
Based on essays and our own research · The book on Amazon Australia
Where this episode comes from.
Hachette, publisher page · hachettebookgroup.com
https://www.hachettebookgroup.com/titles/sherry-turkle/artificial-intimacy/9780316573962/
Peter Dizikes, "Who we become when we talk to machines", MIT News, 29 Sep 2026 · news.mit.edu
https://news.mit.edu/2026/when-we-talk-to-machines-sherry-turkle-book-0929
Samantha Laine Perfas, "My chatbot is my boyfriend. What's the harm?", Harvard Gazette, 14 Oct 2026 · news.harvard.edu
https://news.harvard.edu/gazette/story/2026/10/my-chatbot-is-my-boyfriend-whats-the-harm/
Hope Reese, Boston Globe Ideas, 28 Sep 2026 · bostonglobe.com
https://www.bostonglobe.com/2026/09/28/opinion/sherry-turkle-book-ai-chatbot/
Steven Levy, "Maybe chatbots shouldn't get to say I or me", Wired · wired.com
https://www.wired.com/story/maybe-chatbots-shouldnt-get-to-say-i-or-me/
James Gleick, "Talk to It", New York Review of Books, 5 Nov 2026 · nybooks.com
https://www.nybooks.com/articles/2026/11/05/talk-to-it-chatbots-artificial-intimacy-inventing-eliza/
Miles Raymer, "SNQ: Sherry Turkle's Artificial Intimacy", words and dirt, 16 Oct 2026 · words-and-dirt.com
https://www.words-and-dirt.com/words/snq-sherry-turkles-artificial-intimacy/
Becky Hogge, Financial Times, Oct 2026 · sherryturkle.com
https://www.ft.com/ (copy on Turkle's site: https://www.sherryturkle.com/reviews-ai)
Kirkus Reviews (starred) · kirkusreviews.com
https://www.kirkusreviews.com/book-reviews/sherry-turkle/artificial-intimacy/
Common Sense Media, Talk, Trust, and Trade-Offs, 16 Jul 2025 · commonsensemedia.org
https://www.commonsensemedia.org/research/talk-trust-and-trade-offs-how-and-why-teens-use-ai-companions
Esther Perel, Where Should We Begin?, "My AI Loves Me Better Than Anyone Ever Could", Mar 2026 · estherperel.com
https://www.estherperel.com/podcast

Based on essays and our own research, checked on 21 October 2026. The author's argument is paraphrased, not reproduced. Check the original before quoting.

Our fact check · 39 claims: 14 confirmed, 21 reported, 2 unverified

Before we record, every claim the hosts make is checked against the sources (this table was last updated 21 October 2026). Confirmed means a primary source or two independent reports; Reported means one party says so; Partly right, Disputed, Unverified and Not yet known mean what they say, and the hosts say so on air. This is the table as we left it.

  1. ConfirmedUS Little, Brown, 29 Sep 2026, 288 pp; UK Bridge Street Press/Little, Brown £22Hachette; Little, Brown UK; FT; Some retailers say 22 Sep or 432 pp; ignore
  2. ConfirmedAmazon AU dp/0349136912 is this bookAmazon.com.au, 21 Oct 2026
  3. ConfirmedProfessor of social studies of science and technology, MIT; founding director, Initiative on Technology and SelfMIT STS; Hachette
  4. ConfirmedJoint Harvard doctorate in sociology and personality psychology; licensed clinical psychologistMIT STS; Hachette
  5. ConfirmedSecond Self 1984; Alone Together 2011; Reclaiming Conversation 2015; Empathy Diaries memoirMIT News; MIT STS; PRH; Hachette
  6. ReportedAbout four decades of researchKirkus; FT: "almost half a century"
  7. Reported"Inner history of technology": what it does to peopleMIT News; Interview
  8. ReportedArc: each other, through machines, only to machinesHarvard Gazette; Interview
  9. ConfirmedAlone Together/Reclaiming Conversation about texting and social mediaMIT News
  10. ReportedDefinition of artificial intimacy (claims to care; substitute)Kirkus quoting book; publisher
  11. ReportedPretend empathy; "pretend empathy is empathy enough"MIT News; publisher copy; Publisher quote used
  12. ReportedMaria and the graduate-school reassuranceFT (Hogge)
  13. ReportedGrowth needs relationships with responsibility and consequence; attachment and empathy need frictionFT; Globe; Paraphrased
  14. ReportedOwen, ninth grader, rehearses with ChatGPTFT
  15. ReportedPractice is a "myth"; total acceptance makes chatbots the "endgame"Wired via MIT News clip; Wired page not reachable
  16. ReportedAttention economy → attachment economyScientific American via MIT News; FT quotes book
  17. Reported"I" implies a self; nothing cares whether you have eaten or are safeHarvard Gazette; Paraphrased
  18. Reported"The original sin of artificial intimacy"Gleick (NYRB); Globe; Only book quotation used
  19. ConfirmedWired headline "Maybe chatbots shouldn't get to say I or me"Wired URL; MIT News; Research pack only
  20. ReportedKeep AI "in its lane"; tool not partnerRaymer (p. 41)
  21. ReportedKeep it out of education and psychotherapyGlobe
  22. ReportedLife stages: childhood, adolescence, love, work, deathFT; MIT News
  23. ReportedGrief avatars can cost us the ability to mournMIT News; WaPo essay
  24. ReportedGrieving mother; daughter no longer visits her dreamsFT; Jang Ji-sung, not named on air
  25. Reported"State of emergency"Globe (she writes); Slate via MIT News
  26. ReportedEmergency because empathy and relationships make us humanSlate via MIT News
  27. ReportedAntonio and Astrid; can't ghost, cheat or leave; Perel's stepping-stone questionWhere Should We Begin?, Mar 2026 (vault write-up; episode public); Public podcast, attributed
  28. ConfirmedGleick: "She is right to worry," especially about childrenNYRB; Review text
  29. ConfirmedGleick: most users well satisfied; "I feel heard"; Turkle cites studies rating companions more compassionate than counsellorsNYRB; Review text; Studies not identified
  30. ConfirmedRaymer: 6/10; "too one-sided"; condescending; relational pluralism; praises vignettes and "Reclaiming the Real"Words and dirt; Review text
  31. ConfirmedHogge: remedies "frustratingly tentative"FT; Review text
  32. ConfirmedRetail copy "over 70% of teens ... rely on AI"Hachette/retail listings; Copy exists
  33. ConfirmedCommon Sense 2025: ~1,060 teens; 72% ever; 33% social/relationships; 12% emotional/mental health; conflation caveatCommon Sense report PDF
  34. UnverifiedAdult "nearly one-third" source—; Not on air
  35. Strong case = long-run prediction, harder to testHosts; Opinion; Said as such
  36. Organisational design choices (I, persona, metrics, lane, hand-off, owner)Hosts; Extension; Said as "our extension"
  37. ConfirmedDrage, MYT 8: "stop catastrophising"; harm from human choices and business modelsDrage book subtitle; drage-what-if-we-got-ai-right
  38. ReportedEzra Klein appearanceMIT News; Not on air
  39. UnverifiedNYT review by Jennifer Szalai—; Not on air

Want to know more?

Sign up for our podcast at agile-insights.com.au.

Read the transcript · Part 1: Primer

Linda Before we start, a short note. This episode talks about loneliness, grief and mental health. If that is hard for you right now, the show notes list services that can help, and in Australia you can call Lifeline on thirteen eleven fourteen. Here is the idea for today. When chatbots offer themselves as friends, lovers and therapists, they teach us to accept a performance of empathy. And that slowly wears away what real relationships need. That is the argument of Artificial Intimacy, by Sherry Turkle. This is Making You Think, episode nine.

Leonard And she has been watching this for a long time. Sherry Turkle is a professor of the social studies of science and technology at M I T, and the founding director of its Initiative on Technology and Self. She has a joint doctorate from Harvard in sociology and personality psychology, and she is a licensed clinical psychologist. Her earlier books include The Second Self in 1984, Alone Together in 2011, Reclaiming Conversation in 2015, and a memoir, The Empathy Diaries. Artificial Intimacy came out from Little, Brown in September 2026, with starred reviews in Kirkus and Publishers Weekly.

Linda She calls her life's work the inner history of technology. Not just what a machine does, but what it does to people. And she puts the arc of her books simply. First we talked to each other. Then we talked to each other through machines, which is the world of Alone Together and Reclaiming Conversation, about texting and social media. Now, she says... we are starting to talk only to machines. That is the first step of the argument. This book is about the third stage, and she thinks it is different in kind.

Leonard So what actually changes when the other side of the conversation is a machine, and not a person I happen to be texting? On my screen, the two can look almost exactly the same.

Linda The machine starts to offer care. Artificial intimacy is her name for A I that claims to care for us and offers itself as a substitute for people, as a friend, a companion, even a lover. What it gives, she says, is pretend empathy. The words of care are there, but nobody behind them cares. As the publisher puts it, the danger is that we come to believe "pretend empathy is empathy enough". The Financial Times review gives an example from the book. An anxious student called Maria is told by Chat G P T that, even if she does not get into graduate school, everything will be okay. It sounds kind...

Leonard It really does.

Linda But nothing is at stake for the machine.

Leonard That is the part I find hard, because it does sound kind. If I come away feeling heard, why should it matter that nobody was there?

Linda Because, she argues, the uncomfortable parts of a relationship are doing work. We grow alongside other people, in relationships with responsibility and consequences. Attachment and empathy both need friction, the rub of someone whose wants are not yours. A chatbot takes the friction away. The same review describes Owen, a ninth grader, who rehearses tricky conversations with Chat G P T until he is sure he will not embarrass himself. People call that practice. In an interview with Wired, Turkle called it a myth, because total acceptance turns the rehearsal into the destination. And there is money in that. Social media, she says, built an attention economy. Chatbots are going after an attachment economy.

Leonard Of course they are. And where does the machine saying I come into it? That sounds like a small thing to build a whole argument on. Plenty of apps on my phone talk to me that way.

Linda She thinks it is the centre of the problem. When a chatbot says I care, or I am here for you, it implies a self doing the caring. In her view there is no such self. Nothing in it cares whether you have eaten, or whether you are safe. She calls that first person voice "the original sin of artificial intimacy". Her remedy is to keep A I in its lane. Use it as a tool, but do not let it pretend to be in a relationship with you. In particular, she says... keep it out of education and out of psychotherapy.

Leonard And she follows this through a whole life, as I understand it. At which stages of life is she most worried about what this does to us?

Linda The book moves through childhood and adolescence, then love, work and death. With children, she sees the line between people and things starting to blur. One eight year old uses the same phone to talk to grandparents and to Chat G P T, and treats them all as things you reach on your phone. At the other end of life are grief bots, chatbot versions of dead relatives. Turkle warns that keeping the dead talking can cost us the ability to mourn. The Financial Times review mentions a grieving mother who met a simulation of her dead daughter. Since then... the child no longer visits her dreams. All of this is why Turkle writes that we are in a state of emergency.

Leonard That is strong language, and some listeners will bristle at it. She explained it in an interview with Slate. Empathy and our capacity for relationships with other people are what make us human. So, before Part 2, a question for you. Think of a conversation with a person that was uncomfortable, and changed you anyway. Would a machine have let that happen?

Linda In Part 2: where her argument is strong, and what her critics say. Then what a famous statistic really measures, and what this means for organisations that put chatbots in front of people. See you there.

Read the transcript · Part 2: Discussion

Linda Before we start, a short note again. This episode talks about loneliness, grief and mental health, and the show notes list services that can help. In Australia, Lifeline is on thirteen eleven fourteen. The idea, in one sentence. When chatbots offer themselves as friends, lovers and therapists, they teach us to accept a performance of empathy, and that slowly wears away what real relationships need. Today, where Sherry Turkle is strong, where her critics push back, and what this means for organisations that deploy chatbots.

Leonard A quick recap of the terms from Part 1. Artificial intimacy is her name for machines that claim to care for us and offer themselves in place of people. Pretend empathy is what they give: the words of care with nobody behind them. Friction is the rub of other people, which she says we need in order to grow. The attachment economy follows the attention economy: what is on sale now is our capacity for attachment. And the chatbot's I, its first person voice, is what she calls the original sin.

Linda Start with the strength. Turkle has studied how people relate to machines for about four decades. Her strongest point is not that people dislike chatbots. It is that what a bot leaves out, like waiting, being misread and being disagreed with, is where we learn. You can hear it in a public case on Esther Perel's podcast, Where Should We Begin? A man called Antonio built a chatbot companion he named Astrid. He valued her because she could not ghost him, cheat on him, or leave without an explanation...

Leonard Which is quite a list.

Linda Perel left the question open. Was Astrid a stepping stone back towards people, or a replacement that removed the need to try?

Leonard Turkle's answer, for most cases, is replacement. Care that can never leave you is also care that never asks anything of you. And her worry about children is the hardest to wave away. A child who treats grandparents and a chatbot as the same kind of thing in a phone is learning something about people very early. That is our reading, but here her argument is at its most persuasive. James Gleick, reviewing the book for the New York Review of Books, writes, "She is right to worry," especially about children.

Linda But Gleick adds a complication that matters. On the book's own anecdotes, he says, most users are well satisfied. I feel heard is a common refrain. And Turkle herself cites studies in which patients rate artificial companions as more compassionate than human counsellors. So her case cannot rest on people complaining. It rests on a prediction about what these habits do to us over years. That is harder to test. Miles Raymer, who reviews books on his blog Words and Dirt, is tougher... He gave it six out of ten, and calls it too one sided, in a way that undermines her credibility.

Leonard His sharpest point is about the people she interviews. He thinks she reads their lives as cautionary tales, as if she knows them better than they do. He calls that condescending. He also says she dismisses too quickly the idea that A I could add to our relationships rather than compete with them. And I have some sympathy. If a lonely person says a chatbot helps, should an outsider overrule them?

Linda Fair question.

Leonard Still, Raymer praises her interviews and her closing chapter, Reclaiming the Real. And Becky Hogge, in the Financial Times, found her remedies "frustratingly tentative".

Linda I would defend her a little there. Turkle is a clinical psychologist, and part of that job is noticing what someone cannot yet see about themselves. Saying a comfort may cost you later is not the same as looking down on you. But I think you are both right in one way. Take people's satisfaction seriously, as Gleick does. Then ask Perel's question of each case. Does this companion send the person back towards other people, or away from them? That, at least, you can look for.

Leonard I can live with that. One more piece of evidence needs care. The retail copy for the book says over seventy percent of teens rely on A I for companionship and emotional support. That overstates the survey it most likely comes from. Common Sense Media surveyed just over a thousand American teens in 2025. Seventy two percent had ever used an A I companion. About a third used one for social interaction or relationships, and twelve percent for emotional or mental health support. The report itself warns that some teens may have counted ordinary chatbot use. Ever used is not the same as relying on. The book's own notes may cite something else.

Linda Now the A I connection, and we should say what is whose. Turkle's proposals are about people and society. Do not let the machine say I. Keep A I in its lane, as a tool, and out of education and psychotherapy. Her worry about the attachment economy is about business models. What follows is our extension, for organisations that put conversational A I in front of customers or staff. A customer service bot, an H R assistant, a wellbeing app, an internal helper. Each is a set of design choices somebody made. Does it say I? Does it have a name and a personality? Is it measured on how long people keep talking, or on whether their problem was solved?

Leonard And the lane matters at work. An H R assistant built to answer leave questions can drift into being where an employee brings a breakdown at eleven at night. Who decided it could? A wellbeing app needs a clear point where it hands the person to a human. And each of these needs a named person accountable for how it treats people. In our last episode, Eleanor Drage told us to stop catastrophising about A I. Turkle runs at the opposite temperature, but the diagnosis is similar: the harm comes from human choices and business models. So a question for you. Which chatbot in your organisation is rewarded for keeping people talking?

Linda Sit with that one. Two short activities, each five to ten minutes, with a pen and paper. The first is called Where the friction went. Write down five conversations from the past week that mattered to you, with people or with machines. Beside each one, note who or what it was with. Then note any friction: a disagreement, a wait, a risk of being misunderstood. Finally, tick the ones that changed your mind about something, even slightly. Look at where the ticks fall. Turkle would expect them to sit beside the friction. You may find she is right, or you may not.

Leonard The second is called The lane test. Pick one chatbot you use, or one your team runs or is planning. Write one line on the job it is for. That is its lane. Then answer four questions. Does it say I, or have a name and feelings? What is it measured on? When does it hand you to a person? And who would answer for it if it hurt someone? If an answer is blank, write one change you would make, or one question you would ask the people who run it.

Linda So, three things to carry away. First, a performance of empathy can feel exactly like the real thing, which is why satisfied users do not settle the question. Second, friction is not a defect in relationships. It is where much of the growing happens. Third, for anyone deploying chatbots, the I, the metrics and the lane are design choices, and each needs an owner. You do not have to share her sense of emergency to take those three seriously.

Linda If you'd like to know more, sign up for our podcast at agile-insights.com.au.

Leonard Thanks for listening.

More on #people

Episode 34 · Primer and discussionWhen the test stops telling you anything: hunger beats tenureEpisode 32 · Primer and discussionAgentic video is real: what to try, and what to checkEpisode 31 · Primer and discussionSpeeding up planning decisions with AI: who checks the checker?Making You Think 12 · Primer and discussionThe Game Changer, by Jason FoxEpisode 13 · Primer and discussionFairness under pressure: a fair AI decision has to hold when someone leans on itMaking You Think 8 · Primer and discussionWhat If We Got AI Right?, by Eleanor Drage

Every episode tagged #people

All of Making You Think (13) · Follow Making You Think in Apple Podcasts or any podcast app

Everything from Agile Insights: briefings, Making You Think and the reference series