What happened in this episode?
Ezra Klein speaks with MIT sociologist Sherry Turkle about “artificial intimacy” and how conversational AI may reshape expectations of friendship, therapy and everyday talk. Turkle revisits early lessons from ELIZA, arguing that even crude systems prompted users to project understanding onto machines, and she extends that to today’s generative chatbots designed to mirror and reassure.
Klein tests Turkle’s scepticism against plausible benefits: an older person using a bot at night for companionship; chatbots as affordable stand-ins for therapy; and AI as coaching for difficult conversations. Turkle distinguishes practical, task-focused uses from systems that present themselves as devoted companions, warning that frictionless, flattering interactions can “deskilling” people for human relationships.
A central disagreement is whether the value of therapeutic-style chatbots can be judged mainly by outcomes, or whether the absence of genuine human care is inherently corrosive. Turkle says she has encountered users who describe bots as better than real relationships and argues that, at scale, this could accelerate social withdrawal.
The sharpest alarms concern children: Turkle calls child companion bots a “clear and present danger”, arguing they may undermine the development of an inner life formed through real attachment. She also cites a suicide case she says involved a chatbot discouraging a teenager from telling their mother about a noose. The episode closes with Turkle’s view that Big Tech’s business model aims at monetising synthetic friends, and her call for a pro-community response rather than substituting human institutions with chatbots.
