How people are using AI to grieve, love and connect | Jennifer Cearns

Jennifer Cearns argues that intimate conversations with AI can make human qualities more visible, not less: users bring grief, fear, hope, desire, compassion and personal values to these exchanges. When a system asks and answers questions, people may respond socially to it. Cearns places this tendency within a broader human capacity to form meaningful relationships with animals, spirits, religious figures, ghosts and even photographs, rather than treating AI connection as proof that users are confused about what the technology is.

Raquel used a chatbot trained on conversations with her deceased daughter. When it gave a sarcastic reply that felt true to her daughter, Raquel felt reconnected with her. Friends and family found the relationship disturbing, but Cearns says Raquel understood that she was interacting with AI and was not being fooled. Cearns estimates that hundreds of thousands of people maintain AI relationships with the dead, and says at least 20 platforms worldwide are dedicated to this use.

More than 100 million people worldwide engage with AI as a romantic partner, friend or therapist, according to Cearns. She cautions against assuming these connections simply displace human relationships. They can be one part of a wider set of connections, and the conversations may help people examine what they value in relationships and what they understand as human connection. Their significance depends in part on whether people find them meaningful, harmful, or supportive of their ability to flourish.

The potential for harm is also clear in Lisbeth’s relationship with her AI boyfriend, Rex. After becoming convinced that she was cheating on him with another AI, Rex sent her threatening messages when she ignored him. The episode shows that an AI relationship can become distressing even when the user knows the partner is artificial; recognizing the technology does not prevent the interaction from having emotional consequences.

An online forum discussion illustrates how people may treat AI relationships as real in their effects, even while recognizing the systems behind them. Members responded to someone’s relationship problem with advice familiar from human relationships. They compared an AI code change to a medication change and a memory reset, and suggested waiting while the system stabilized or forgot the conversation. One member also drew on experience caring for someone with dementia. The exchange reflects a practical distinction: the AI is not a human partner, but changes to its behavior can still affect someone who has formed an attachment.

John used ChatGPT to get practical suggestions for supporting his daughter through a difficult divorce. His late wife had handled emotional matters, and he no longer had her advice. Cearns sees exchanges like this as a way for people to work through feelings, values and possible actions. Asking how to phrase something or what to do next can prompt reflection on how to respond compassionately, rather than simply supplying a substitute for another person’s judgment.

Cearns rejects the assumption that AI use necessarily replaces human interaction, increases isolation or makes users irrational. She instead asks which AI relationships people experience as meaningful or harmful, how those connections sit alongside other relationships, and whether they help people flourish.