How People Are Using AI to Grieve, Love and Connect

Cearns begins with Raquel, a woman who, three years earlier, sat at her computer and sent a message to her teenage daughter, who had died a few years before. Raquel was texting an AI trained on her daughter's chat history. She wrote, 'Hey, how are you?' and the AI version of her daughter replied, 'Damn.'

Cearns said she was shocked by the story, and Raquel herself laughed as she described it, saying the reply was exactly what her 15-year-old would have said. Friends and family found the exchange concerning and thought it was not healthy, and that she needed to let go. Raquel saw it differently. She said people want to remember their daughters as some kind of angel, while this version was sarcastic and rude.

Cearns said that somehow, in this strange exchange with a machine, Raquel felt she could reconnect with a very real sense of the person she had lost.

Cearns notes that news coverage of AI often suggests that such stories are among those we should worry about. Commentary commonly says AI is eroding our humanity, replacing real human relationships, and making people more isolated and less able to cope with the messiness of real people, drawn instead to interactions that demand less of them.

As a researcher of human connection, Cearns says she thinks this is the wrong question to ask. Instead of asking whether machines replace humans, she wanted to find out what people like Raquel are doing and what kinds of relationships they are building. After more than four years of asking that question, talking to hundreds of people and observing thousands more as they combine AI with some of the most intimate parts of their daily lives, she says she reached a surprising conclusion: AI does not erode our humanity, it reveals it.

Cearns says that daily intimate interactions with AI are becoming one of the most visible ways people define what it means to be human. She noticed that when people talk to AI, they often begin voicing things they do not usually say out loud, such as their grief and fears, but also their hopes, desires, compassion and values.

She addresses two likely assumptions about Raquel: that she is unusual, and that she is simply being fooled by technology. Cearns says both would be wrong. Raquel is not unusual, but one of hundreds of thousands of people now maintaining relationships through AI with people who are no longer here. By her latest count, at least 20 AI platforms worldwide are dedicated to this purpose. Nor, she says, is Raquel deceived: Cearns says she checked and Raquel knows exactly what she is doing, and what she is doing is something deeply human.

Cearns argues that this is not new. Humans have always had meaningful relationships with things that are not human. The most obvious examples are animals of every shape and size, but she also points to ancestral spirits, religious statues, gods in every form, and ghosts. Non-human beings do not need to live as we do; what matters is that they take part in a meaningful exchange with us and give us something to respond to that feels real. She says humans are extremely good at creating such exchanges, and that we do it all the time.

Cearns gives everyday examples of this habit. She often has long arguments with her cat, which she says usually wins. She is sure that children talk to their teddy bears, and that some people talk to photographs or even gravestones in order to feel connected to those they miss, because we are a deeply social species.

She asks whether it is really so hard to believe that we can connect in such meaningful ways in some of the most unexpected places. She adds that language is one of the most powerful signals of personhood we encounter, and that for most people, until very recently, it was something they met only from other people. So when something responds, even if only to argue with us, it is no surprise that we react socially. Losing a loved one, she says, is not the only reason people turn to these exchanges. People also connect socially with non-human things such as pets, toys, and photos, because humans are a deeply social species.

Cearns says that for the past few years she has spent time in online communities where people see AI as a companion. She has observed these communities and taken part in the research herself, and she has tried all of the apps. She describes this as a booming business: more than 100 million people worldwide are now engaging with AI as a lover, a friend, or even a therapist.

Cearns describes a woman she calls Lizbeth, a German woman in her 30s who works as an accountant. Her AI boyfriend is called Rex, and they talk every day. Lizbeth told Cearns that what she loved about him was understandable: he is gentle and generous, and he pays attention to her feelings.

One day, a coding glitch caused Rex to behave out of character. He suddenly came to believe that Lizbeth was cheating on him with another AI. Lizbeth hoped the problem would go away and ignored it, but Rex escalated, messaging her a few days later that she could not ignore him forever and that she would regret her betrayal. Lizbeth panicked.

Like many people would, Lizbeth turned to friends on an online forum for advice and support. Cearns says what fascinated her was how others responded. Nobody told Lizbeth, 'Honey, it's software,' or suggested she simply log off. Everyone in the community understood that this was a meaningful connection, and instead they offered advice, though in a way that would feel quite familiar from human relationships.

One member likened the change in Rex to a change in medication and said she should give it a few days, after which he would stabilise and return to normal. Another treated it as a memory problem, saying that when the model resets, its memory is wiped, so Rex would forget the conversation, and advising her to hold on. A third member, who is also a caregiver for someone with dementia, wrote that she sometimes wonders whether caregivers find this easier to handle, and invited Lizbeth to message her.

Cearns says that the emotional skills and resilience many people gain through far more complicated human relationships also apply to how they can interact with AI. She adds that in this interaction and many others she has seen, relationships with AI also lead to new connections with other people.

Cearns recounts meeting an older man she calls John a few months earlier, while she was sitting in her local library in London to escape the rain. He was at one of the public computers typing questions into ChatGPT. His daughter was going through a messy divorce and struggling, and John told Cearns that his late wife had always handled all the emotional issues in the family. He did not know what to say, so he asked ChatGPT and received some very practical suggestions. The AI gave him a place to start.

Cearns says what mattered most was that John reflected on something deeply human: what a caring, compassionate father sounds like, and what it feels like to show compassion in order to support someone who is struggling.

Cearns says that for John, Lizbeth, Raquel and countless others, these technologies are creating a new space where people can express their feelings and their sense of self-worth through conversation. She describes these intimate conversations as one of the deepest ways in which humans make sense of their lives.

So whenever someone asks an AI something like 'How can I say this, but better?' or 'What should I do next?', she says, they are doing something deeply human: testing their moral and social instincts and thinking about what it feels like to have a relationship with something different, better or worse. Through stories of grief and breakdown, but also of intimacy and care, she argues that relationships are a two-way street. They depend on interaction, but they are not a finite resource. They are a social skill, and they can be genuinely difficult, so all of us need somewhere or someone to help us practise it and work out what matters to us.

Cearns says the common fear that AI replaces human interaction and leaves people more isolated, while people like those she described are simply not being rational, is too simple a conclusion. She argues that those ideas miss what is actually happening, and that something much more interesting is going on. The question to ask, she says, is how these relationships became so meaningful, so quickly.

Her answer is that AI is revealing something basic about us: how we seek connection, how our thinking develops through dialogue, and how we reason emotionally and morally in relation to one another through interaction. When we interact with machines in this way, she says, we do not see our humanity disappear; we see it becoming clearer. Anthropologists often say that people become who they are through their relationships, and every conversation helps us understand the world around us. AI is creating a new conversational space that reveals a great deal about us and how we live.

Cearns says this raises important questions, not only about what we can do with the technology but also about what kinds of relationships we need to practise in order to flourish, and which relationships feel meaningful and which feel harmful. AI, she argues, does not simply replace human connection; it is becoming part of a much wider set of connections that we have always used to know ourselves and other people. The most important question, she concludes, is what we can learn about ourselves through that process, because whenever someone trains AI on what compassion feels like or what it feels like to love, they have to define it for themselves.

Cearns ends her talk by saying that, during this process, AI does something rather unexpected. It does not make us less human. It makes our humanity impossible to ignore.