Why Tech Promised Community and Delivered Chatbot Friends

Sherry Turkle, the MIT sociologist who has spent four decades studying how people live with machines, sat down with Sam Harris to discuss her new book, Artificial Intimacy: Who We Become When We Talk to Machines. The conversation moves from the utopian early internet to the attention economy, from the disappearing habits of solitude and boredom to the question that gives the book its title: whether a program that performs empathy can ever be enough for a lonely person. Turkle's answer is a firm no, and her reasoning runs through a story about the people who built the first personal computers and the slogan they chose for their future. What follows draws on the opening portion of that exchange and the episode's published description, and it lays out the argument Turkle makes about where we went wrong and why she thinks the current pitch for AI companionship is the wrong fix.

What did the first personal computer people actually want?

Turkle's story begins in San Francisco in the 1970s, when she was visiting the early personal computer owners. She remembers a group called Community Memory, started by Lee Felsenstein, a pioneer of the personal computer movement. The group's mantra, as she recalled it to Harris, was short and almost naive in its hope: use the virtual to improve the quality of life in the real.

That was not a marketing line for a product. It was a theory of what computers were for. The idea was that a networked screen would be a neighborhood hub, a way to find people nearby, trade information, and then go out and meet them. Turkle was an early sympathizer. She studied the WELL, the Bay Area online community, though she was never an active user, and she spent time with its members. The aim, she says, was to create community online that would translate into more and better community offline.

Harris, who is roughly her contemporary in the culture of that moment, admitted to a nostalgia for it. He described listening to early online adopters from around 1994, including the writer Terence McKenna, whose predictions about culture moving into cyberspace were, in his words, quite prescient even if the optimism now sounds quaint. Harris's question to Turkle was the right one for the period: what could possibly be wrong with more access to the world's information and less friction in connecting people?

Why isn't technology ever good or bad by itself?

Turkle's first move is to challenge the word "intrinsic." It is the word that lets technologists off the hook. A tool, the argument goes, is simply good for connection, and misuse is an afterthought. Turkle rejects that framing.

"I think the notion the problem there is the word intrinsic. With technology nothing really is intrinsic because right away the social the political the economic environment in which the technology is being shaped jumps in."

She illustrates the point with Facebook, which she first encountered as a promise. Grandmothers looking at pictures of grandchildren. Classmates from all over the world holding a kindergarten reunion. Turkle describes having a fifth-grade reunion with the people she went to school with, something she calls unthinkable in an earlier era. There was every reason, in that moment, to feel that connecting the world was good in itself.

Then the business model arrived. Turkle's point is that the same architecture could be tuned for a different goal. If attention becomes the most important thing about a platform, its designers have a strong incentive to keep people staring at their screens. The method, as she lays it out, is grim and specific: make people angry, then keep them with their own kind, siloed into groups that agree with them. That keeps eyes on the feed, which makes it possible to sell ads and scrape data. The result is profitable for the platform and corrosive for the person who is angry, anxious, activated all the time, and not going out to meet people.

What is lost when solitude and boredom disappear?

The attention economy does more than fragment public life. Turkle argues it erodes the private capacities that make real relationships possible. In her account, people who are always reachable lose the experience of being alone with their own thoughts, and with that they lose the ability to be comfortable with themselves. Her 2011 book Alone Together, which she references as part of her body of work, made a similar case about tethered life; the new book takes the argument further, into a world where the screen can talk back.

This is where the stakes get personal. If a person can't tolerate the friction of a difficult friend, a rough conversation, or the silence of a long wait, the easier option is always there. Turkle's concern is not that people will consciously choose machines over friends. It is that the choice will be made gradually, one frictionless exchange at a time, until the harder relationships feel like too much.

Why did Turkle say we should not give up on the real?

The most direct passage in the opening exchange is Turkle's response to the current industry pitch. She describes Mark Zuckerberg's recent remarks about Meta, which she paraphrases as saying that people would like to have fifteen friends but that this is unrealistic because they are too busy with jobs. The proposed fix, as she tells it, is to supply the missing twelve or thirteen friends as avatars, programs, and chatbots tuned to each user, while the person keeps two or three real ones.

Turkle treats this as a telling admission. The corporate answer to a loneliness crisis is to stop expecting people to join their communities at all. Instead of rebuilding the neighborhood hub the early computer people imagined, the proposal is to replace the neighbors with screens. She sees that as a shift from the optimistic slogan to its inversion.

"Don't abandon the real now. You know, reality really needs you. This is not a good time to abandon the real."

She is blunt about the costs. Withdrawing into chatbot friendships, in her view, is a disaster for the individual, for communities, for children as they grow up, and for politics. The argument is that democratic life depends on people who can tolerate disagreement and vulnerability with other humans, and an economy that rewards retreat from exactly those skills works against it.

Can a machine performing empathy ever be enough?

The emotional center of the conversation is empathy. The clip that opens the episode carries Turkle's sharpest formulation of the problem: that for many people, the performance of empathy by a program becomes empathy enough. She frames it as the problem she believes we face today, one that is larger than any single product.

Harris and Turkle spend time on what empathy actually is. The distinction she draws is between a response that signals attention and an experience that involves another mind. A chatbot can produce the signals with remarkable fluency. Turkle's argument is that signaling care and caring are different things, and that a person who gets only the first may mistake it for the second. The discussion also touches on research in empathy, including the work of psychologist Sara Konrath and the development of empathy-focused apps, which the speakers treat as a sign of how seriously the culture now takes the question.

The book's title points at the core issue. Artificial intimacy is intimacy without the costs and reciprocity of a relationship with another being who can be hurt, disappointed, or changed by you. Turkle's concern is that those costs are part of what makes a relationship meaningful, and that removing them is not a neutral upgrade.

Do humanoid robots and AI systems actually care?

The later part of the conversation turns to whether machines can care, and the description of the episode lists humanoid robots, consciousness, and the role of AI as therapist among the topics covered. Harris and Turkle approach the question with interest and some skepticism. A machine can be built to behave as though it cares, and that behavior can genuinely comfort a person. The open question, which the conversation keeps circling, is whether the comfort is grounded in anything on the other side of the exchange.

The conversation also takes up loneliness among older people, a group for whom AI companionship is often offered as a solution. Turkle's broader position, consistent with the rest of the talk, is that the answer to isolation should be more human contact rather than a substitute that asks less of everyone involved.

Where to go from here

The most useful way to approach Turkle's argument is to begin with the history she keeps returning to. The early personal computer people believed the virtual could improve the real. Turkle's work asks what happened to that belief, and whether it can be recovered. Her answer is not a rejection of technology. It is an insistence that we measure tools by the human relationships they strengthen, rather than the engagement they capture.

For readers who want the full argument, Artificial Intimacy: Who We Become When We Talk to Machines is the place to start. Turkle's earlier books, Alone Together and Reclaiming Conversation, supply the longer history of the same concerns. Her website, sherryturkle.com, lists her current work. And Harris's full conversation with her is available through his podcast subscription on his site, where the rest of the discussion, including the topics the opening clip only hints at, is developed in full.

The question Turkle leaves behind is practical. Before accepting a machine as a friend, ask what the relationship asks of you and what it asks of the other party. If the answer is nothing, the comfort may be real, but the friendship is not what it appears to be.

The complete publication date of this conversation is listed on the source video as November 3, 2026.