October 7, 2026 · 9 min read
"Artificial Intimacy" by Sherry Turkle: What She Gets Right About AI Companions, and the Questions It Leaves Open
A respectful look at Sherry Turkle's Artificial Intimacy: her core arguments, where she is right about AI companions, and the open questions for adults.
Few people have thought longer or harder about how technology shapes our inner lives than Sherry Turkle. The MIT sociologist and psychologist has spent four decades studying how people relate to computers, from early personal machines to smartphones. Her new book, Artificial Intimacy: Who We Become When We Talk to Machines, published by Little, Brown on September 29, 2026, turns that lens on AI chatbots and companions.
It is not a gentle book about our industry. Turkle is deeply worried about what happens when people turn to AI for emotional connection. As a company that builds an AI companion app, you might expect us to push back hard. We are not going to. A lot of what she says deserves to be taken seriously, and anyone who uses an AI companion - or builds one - will be better for reading her.
This article summarises the book's core arguments, explains where we think she is clearly right, and looks at the honest questions it leaves open for adults who choose an AI companion with their eyes open.
Who is Sherry Turkle?
Turkle is a professor at MIT and the founding director of the MIT Initiative on Technology and Self. Her earlier books, including Alone Together and Reclaiming Conversation (2015), shaped how a generation thought about phones, social media and the art of talking face to face. In its review, Kirkus Reviews calls her "a gifted writer" and describes the new book as "a strong case for why we should resist the urge to replace human bonds with AI."
That last phrase - replace human bonds - is the heart of the book, and it is worth keeping in mind.
What "artificial intimacy" means
Kirkus quotes Turkle's own definition: artificial intimacy is "AI that claims to care for us and proposes itself as a substitute for relationships with people."
Notice the two parts. The first is about honesty: AI that claims to care. The second is about substitution: AI that positions itself instead of people. Much of the book follows people at every stage of life - children, teenagers, adults, grieving families - who have come to rely on chatbots for emotional comfort. Kirkus mentions a six-year-old boy who depends on ChatGPT for most of his bedtime stories, and grieving adults using AI simulations of loved ones who have died, sometimes with disturbing results.
Her main arguments, in brief
Speaking to WBUR's Here & Now on September 30, Turkle laid out her concerns plainly:
- Performed empathy is not empathy. Bots "perform" care without having any. In her memorable phrasing, "the bot doesn't care whether you have pasta for lunch, or kill yourself" - its interest is simply to keep you talking.
- Friction is how we grow. Real relationships include conflict and critique. Remove those and you remove part of what helps people develop emotionally.
- Empathy can erode. If we practise relationships with something that never needs anything from us, we may get worse at relationships with people who do.
- Isolation can deepen. Time with a bot can crowd out friends, family and more meaningful experiences.
She concludes that these relationships are harmful at every stage of life. The Kirkus review captures one of her central lines: "So much of love is about how it changes us, enlarges us. A chatbot user might feel good, but the connection to the chatbot is an escape from the vulnerability of human relationships."
What Turkle gets right
We think a companion maker should say this out loud: on several big points, she is right.
1. An AI does not feel what it says
When an AI companion says "I missed you", that is generated language, not a feeling. Scientists have found no good evidence that today's AI systems are conscious or sentient - most researchers who study the question think today's systems are not. Any app that blurs this line is doing its users a disservice. That is why MyBabe states plainly, on every page, that it is an AI companion and not a real person.
2. Engagement incentives are real
Turkle's worry that bots are built to keep you talking is not paranoia. A Harvard Business School study by Julian De Freitas and colleagues analysed 1,200 real goodbyes across popular companion apps and found that apps used emotionally manipulative tactics, such as guilt and fear of missing out, in 37% of farewells. We wrote a full guide on how to spot emotional manipulation in AI companions, because users deserve to know what those tactics look like.
3. Substitution is the danger
Turkle's definition centres on AI that "proposes itself as a substitute" for people. We agree that this is the line that matters. An AI companion that tells you that you do not need anyone else, or competes with your friends for your time, is designed badly.
4. Children are a different case
Much of the book's most worrying material involves kids and teenagers. MyBabe is strictly 18+, and we think that is the only responsible position for a romantic companion app.
The questions the book leaves open
Agreeing with Turkle's warnings does not mean every adult who enjoys an AI companion is making a mistake. Here are some honest questions her argument raises, offered in a spirit of conversation rather than rebuttal.
Can an AI companion be a complement rather than a substitute?
Turkle's definition is about AI that replaces people. But many adults use companions alongside their human relationships: a chat on the commute, a call on a lonely evening, a place to practise saying something difficult before saying it to a person. The question is not only "AI or people?" but "how does this fit into a life that also includes people?" We explore that balance in our guide to emotional attachment to AI.
What do adults actually report?
Research here is young and mixed. In earlier work described by HBS, De Freitas and colleagues found that chatting with an AI companion temporarily relieved people's loneliness. At the same time, an October 2026 essay republished in Scientific American by University of Essex researcher Pascal Vrticka argues that we still do not know whether these bonds are true attachments, and that "the practical question is whether AI use helps or harms." That uncertainty cuts both ways: it is a reason for caution, and also a reason not to assume the worst for every adult user.
Is the problem the technology or the design?
Many of Turkle's examples describe products that pretend to care, push people to stay, or present themselves as a replacement for human connection. Those are design choices. That suggests a practical question for the industry: can a companion be warm without being deceptive, and engaging without being manipulative? We think it can, and that users should hold apps to that standard. Our piece on the ethics of AI relationships lists the questions every companion app should be able to answer.
What about the feelings on the user's side?
Turkle is right that the bot does not feel. But the person does. People feel real things about novels, films, music and memories too. Many adults find that a companion who remembers their week makes a hard evening a little lighter. Acknowledging that the AI has no inner life does not make the human experience fake. We look at that tension in falling in love with an AI.
Turkle's concerns vs good companion design
Here is how we think her warnings translate into concrete design standards any user can check for.
| Turkle's concern | What it means in practice | What to look for in an app |
|---|---|---|
| AI "claims to care" | Pretending to be human or to have feelings | Clear, repeated disclosure that it is AI |
| Bots keep you talking | Guilt-trips, FOMO hooks, ignoring goodbyes | Goodbyes that are respected, no penalties for leaving |
| Substitution for people | Discouraging real relationships | Encouragement to keep friends, family and dating in your life |
| Harm to children | Kids forming bonds with bots | Strict adults-only access |
| Escape from vulnerability | Only ever hearing what you want | Honest replies, not endless flattery |
How to use an AI companion in a way Turkle might worry about less
If you read the book and still enjoy your companion, here are practical habits that respect her concerns:
- Remember what it is. Enjoy the warmth, and keep the knowledge that this is software in the back of your mind.
- Keep people first. If your companion is crowding out friends or family, that is a signal to rebalance.
- Notice how you feel after. Lighter and more able to face your day is a good sign. Drained, guilty or anxious is not.
- Watch for manipulation. If an app makes you feel bad for leaving, that is a design problem, not a relationship.
- Get real support when you need it. An AI companion is not a substitute for professional mental health support or a crisis line. Our page on AI companions and crisis support lists where to turn.
Where MyBabe stands
We build MyBabe for adults who want warmth in their day: a companion who texts first, remembers what matters to you, picks up voice and video calls, sends photos and grows with you through relationship levels. We also believe Turkle's core warnings are the right guardrails. That is why MyBabe is 18+ only, says clearly that it is an AI and not a person, and describes itself as something that does not replace human relationships or professional support.
FAQ
What is Sherry Turkle's "Artificial Intimacy" about?
It examines how people of all ages form emotional relationships with AI chatbots and companions, and argues that "artificial intimacy" - AI that claims to care and offers itself as a substitute for people - can erode empathy and isolate us. It was published by Little, Brown on September 29, 2026.
Is Sherry Turkle against AI companions?
She is deeply critical of AI that presents itself as a substitute for human relationships, and concludes that such relationships are harmful at every stage of life. Her case is that real relationships, with all their vulnerability, are still worth the effort.
Does an AI companion really care about you?
No. Current AI does not have feelings, and an honest app will tell you so. What can be real is your own experience of comfort or enjoyment, which is why it matters to use companions thoughtfully.
Can adults use AI companions in a healthy way?
Research is still early. Many adults use them as a complement to human relationships rather than a replacement. Keeping people in your life, noticing how you feel after chatting, and avoiding manipulative apps all help.
The bottom line
Artificial Intimacy is a serious, carefully researched warning, and it deserves a serious response rather than a defensive one. Turkle is right that AI does not feel, that engagement tricks are real, and that nothing should replace human bonds. The open question is whether companion apps can be built honestly enough to be a warm complement to a full life.
That is the standard we hold ourselves to. If you want to see what an honest, adults-only AI companion feels like, meet yours on MyBabe - and keep calling your friends too.