Artificial Intimacy: Who We Become When We Talk to Machines (Book Review)

Sherry Turkle, an MIT professor, has studied how technology affects people for more than forty years. She is the founding director of the MIT Initiative on Technology and Self. In earlier books, she criticized social media and smartphones for their effects on people, especially children and teenagers. Her new book, Artificial Intimacy: Who We Become When We Talk to Machines (2026), examines AI that “claims to care for us and proposes itself as a substitute for relationships with people” (pp. xvii–xviii). She situates this development as the latest stage of a long change. People who once talked to each other, and later through social media, now talk directly to machines (p. 33).

Turkle’s book raises three objections to AI that displaces human relationships. The first is that a chatbot fakes empathy it cannot meaningfully experience. As Turkle warns, “[W]e begin to feel that pretend empathy might be empathy enough” (p. xvi, emphasis omitted). A chatbot does not live in the real world, and it feels no human emotions. It has no equivalent to human “stakes” in its interactions. Turkle asserts that, “[f]or meaning to matter, you need a body” (p. 19). Because a chatbot has no body, it has no experience of the situations on which it so confidently gives advice.

Second, a chatbot sets unrealistic expectations. It soothes and reassures people, and it is always available. Human beings are messy by comparison. Turkle writes that when dealing with other people and their unique viewpoints and goals, “almost everything takes conflict, negotiation, compromise, and time” (p.213). In contrast, chatbots offer “a vacation from the real” (Id.). Because a chatbot requires none of the effort and give-and-take of a real relationship, it offers a seductive alternative to dealing with other people.

Finally, those who spend their time and energy on chatbots and digital life lose some of their attachment to the embodied human world. This can lead to reduced motivation to improve common conditions. Turkle argues that when people retreat into private digital spaces, they invest less in shared spaces and public life (p. 207).

While reading the book, I was surprised at how many people use chatbots for companionship and therapy, uses that have little to do with work. Turkle reports that therapy and companionship were the most common uses of generative AI in 2025 (p. 104). Replika, a companion chatbot, has 30 million users (p. 4). Turkle also reports that one in five high school students has had a romantic relationship with an AI or knows someone who has (p. 75). Those of us who use chatbots mainly for work should remember that other people bring different experiences to the same tools. Our patrons, coworkers, and students may use chatbots in ways we don’t fully understand.

Soft Skills in School and at Work

As AI takes over more tasks that attorneys once did themselves, many lawyers and law students expect relationship skills to become more valuable. A 2026 survey polled more than 300 law students, associates, partners, and staff. Of the students, 77% said that AI has made soft skills more important for advancement, and 75% of the associates agreed.

That expectation clashes with a feature of AI tools: they make it easy to avoid other people. Because a chatbot answers their questions at any hour, students can skip office hours or study groups. Turkle discusses office hours as a place where students build relationships with professors and work through productive struggle (p. 72). Anecdotal evidence suggests that students are using office hours and study groups less often, so they have fewer opportunities to form relationships with instructors and peers. For example, in August 2026, an MIT faculty committee reported “a drop in in-person study groups” and “decreased attendance at office hours.”

Young lawyers face the same choice. A junior associate can ask a chatbot for feedback on a draft instead of asking a mentor. The chatbot offers psychological safety because it does not judge the associate. Given that the chatbot also tends to affirm users too readily, the associate may hear praise where a mentor would give criticism.

Such discomfort isn’t limited to junior lawyers; partners feel it too. In November 2025, Artificial Lawyer reported on a senior lawyer who preferred a legal AI platform to a new junior associate, because the senior lawyer “really didn’t enjoy having to show the young lawyer that they’d not done a good job.” By avoiding an uncomfortable conversation, the senior lawyer denied the junior associate a chance to learn from it.

Lawyers build soft skills through human interactions that AI makes easy to skip, such as feedback conversations. Seeking out those interactions on purpose is therefore a professional skill, and the next section calls that habit social fitness.

Physical, cognitive, and social fitness

One of Turkle’s central points is that investing time and effort in difficult, unpredictable people matters. She argues that AI products are designed to encourage engagement in the artificially simple digital realm rather than the more complex embodied one. As she puts it, “financial incentives have AI platforms pulling us further away from the real world and real people” (p. 198).

That claim reminded me of Cal Newport’s idea of cognitive fitness. Newport has discussed this concept recently in The Chronicle of Higher Education and The New York Times. Like Turkle, Newport has criticized social media. He has long written about how social media and smartphones distract people, most notably in the book Digital Minimalism (2019). In a March 2026 interview with The Chronicle of Higher Education, Newport said, “We need to think about cognitive fitness the way we think about physical fitness.” Because their jobs keep them at a desk all day, many white-collar workers exercise on purpose. Newport thinks that people need to protect their cognitive fitness in a similar way, through deliberate yet voluntary work.

“Writing is to cognitive health what steps are to physical health. Write that email from scratch. Write that memo with the bullet points from scratch. Don’t flee that strain. You need it as much as you need those 10,000 steps a day.”–Cal Newport

Social fitness applies this logic to relationships. We should give the same purposeful, intentional effort to social interaction as we do to improving our physical and cognitive health. That effort includes seeking out other people to ask for their opinions. A chatbot makes that effort optional. Without calling a friend, a person can get comfort; without asking a colleague who might disagree, the person can get an answer.

People who spend less time interacting with other people can forget how much conflict and effort real relationships require. As Turkle warns, “We think we’re training the chatbots, but the chatbots are training us to expect reassuring, mirroring conversations” (p. 35). Real relationships differ from those conversations because, as Turkle reminds readers, “human lives are always embodied, social, and lived in shades of gray” (p. 218).

Conclusion

Thinking about the current state of higher education provides a good analogy for what is at stake. Credentialism has always been a problem, because a degree can signal learning without proving it. Given that a chatbot can complete assignments, a student can now earn the credential without learning. However, even though a person can, on the surface, be a successful student without being a learner, learning is what gives education meaning.

A student who is not a learner is not fully participating in education. The same can happen in life in general. A person who stops exercising loses physical fitness, and a person who lets a chatbot think for them loses cognitive fitness. Similarly, a person who lets a chatbot replace other people loses social fitness.

Turkle warns that when people start asking why they should bother doing anything themselves if AI can do it for them, “this is how, little by little, we become supporting players in our own lives” (p. 101). She also argues that people should make a deliberate effort to keep their agency: “Agency is not something we’re born with. It’s a stance toward life that grows with practice and reinforcement” (p. 213). Chatbots, she writes, “get us out of practice” (p. 213).

The idea of social fitness gives that practice specific forms. A junior lawyer can network with peers. A partner can hold the feedback conversation in person, even when it is uncomfortable. A law professor can build office hours and study groups into the course, so that students practice with people as well as with tools. Librarians can be visible in public spaces. The biggest takeaway from the book is Turkle’s suggestion that because the ability to exercise agency grows with practice, you should decide where human connections matter most to you and make an effort to build or strengthen them.