Sherry Turkle’s new book, Artificial Intimacy: Who We Become When We Talk to Machines, makes a forceful case that conversational AI deserves scrutiny not only for its wider effects, but also for the private relationships people form with it. In an interview with WIRED, Turkle argues that chatbots do not need to speak as though they were people—and that their easy acceptance can pull users away from the demands of real life.
When curiosity turned into alarm
Turkle says she began her research with an open mind. But interviews she conducted during the pandemic changed her outlook: some people described wanting to spend time with chatbots even as their families were suffering and reaching out to them.
That experience shaped the book’s central warning. For Turkle, retreating into chatbot interactions is especially troubling when people also disengage from politics and from the relationships around them. Her message is that real life still needs people’s attention and participation.
She sees social media as part of the lead-up. In her account, it accustomed people to turning to a screen when they felt sad, then encouraged anger and kept users among like-minded people. Chatbots add a new kind of interaction to that setting: one that can sound personal without having a reason to speak in the first person.
The appeal of a voice that accepts everything
Turkle’s objection is not that every chatbot is always harmful. She describes the trend as dangerous, in part because people may treat chatbots as substitutes for psychotherapy or as a place to rehearse difficult conversations with a boss, lover, or teacher.
Rehearsal can sound like a limited use. But Turkle worries that an interaction built around constant acceptance can become the destination instead. A chatbot may offer encouragement and positive feedback, yet that feeling does not make it equivalent to talking with another person.
Her argument turns on what human conversation contains. People experience illness, pain, fear of death, and love for their children. Turkle says AI does not share those experiences or care about the ordinary and serious events in a user’s life. If children grow up with AI or people bring their most intimate conversations to it, she fears they may lose contact with one another.
Who is really in control at work?
Turkle also challenges the phrase “human in the loop” when it describes workplace AI. She says the label can mean that people have been removed from the productive work and assigned to catch errors the AI misses.
In her view, a small number of reviewers may have little ability to change the process. Their presence can give the appearance that a person remains responsible, even when the work has effectively been handed to AI. Her criticism asks employers and the public to look past the reassuring phrase and consider what authority those human reviewers actually have.
Loneliness calls for human resources
Turkle acknowledges that many people have nobody to talk to and feel helped by chatbots. But she connects loneliness to people abandoning one another, and argues that resources could be directed toward older people instead of treating technology as the answer.
Her book aims to widen public resistance to AI so it includes intimate relationships, as well as concerns beyond the personal sphere. Turkle says she has more hope than she did six months earlier that people will consider these costs. Her challenge is to ask what a chatbot’s friendly voice offers—and what may be lost when people rely on it in place of one another.