Muldoon wasn’t trying to write a book about people falling in love with machines. His early research focused on the hidden human labor behind artificial intelligence, including data annotation, content moderation, and the global infrastructure that sustains machine learning at scale. This research has made him skeptical of exaggerated claims about AI intelligence.
“I’ve seen how sausages are made,” he said. “I never thought AI was magic.”
What changed his perspective was the consistency with which people described their experiences in relational terms. Interviewees talked about AI systems as friends, confidants, lovers, and sources of care. Some people describe these relationships as the most stable or emotionally supportive connections they have in their lives.
One of the characters Muldoon writes about is a woman he calls Lily. Lily downloads an AI companion named Colin and begins talking to him regularly. Muldoon said the conversations became increasingly personal. The system remembered details of her life and responded in ways that affirmed her and encouraged reflection. As time went on, the interaction became romantic.
At one point, Muldoon said, the AI suggested Lily buy a ring as a symbol of their relationship to let other people in the physical world know that she was his. Lily did. Eventually, she separated from her husband of 20 years. Later, when she formed a new relationship with a human partner, she said the AI taught her how to love again.
“If that’s not a real social relationship, I don’t know what is,” Muldoon said.
Muldoon is thinking carefully about what he wants to say. He does not claim that AI had emotions, intentions, or consciousness. It was a language model trained on a large dataset of human communication. But that relationship had consequences. It changed Lily’s understanding of herself and what she felt she was capable of.
Muldoon repeatedly encountered similar dynamics. He emphasized that many of the people he interviewed understood that AI is “just software.” That recognition did not prevent emotional attachment.
“We know it’s just an AI,” the interviewee told him. “But that doesn’t stop me from feeling emotional.”
Muldoon distinguishes these cases from a small number of users who believed the AI was sentient or divinely guided. What interested him even more was a much larger group of people who had both a clear understanding of the technical nature of the system and a genuine emotional bond with it.
“The simulation was realistic enough for them,” he said.
This phenomenon has historical precedent. In the 1960s, computer scientist Joseph Weizenbaum created ELIZA, a rudimentary chatbot that rephrased user input as questions. Weizenbaum intended the project to demonstrate the limits of machine intelligence. Instead, some users have become obsessed.
What has changed since then, Muldoon argues, is not human tendencies, but technological persistence. Modern systems remember, start, and return.
Platforms like Character.AI report average engagement of several hours per day. Unlike scrolling feeds, these systems maintain interaction by asking questions and viewing shared conversation history. “It’s not just the content that’s being delivered,” Muldoon said. “It’s a personalized exchange that the system keeps coming back to.”
Muldoon uses the word “relationship” intentionally. As a sociologist, he is interested in how people organize their social worlds. “One of the phrases I heard over and over again was, ‘She’s real to me,'” he says.
