This Is Not Just Something Children Do

In Anthropomorphism in Children’s Interactions with LLM Chatbots, Hansinie Madushika Jayathilake and Renkai Ma review how children assign human characteristics, intentions, and emotions to chatbots.

“LLM conversational coherence alone is sufficient to trigger anthropomorphism.”

The paper synthesizes 35 studies and identifies several things that encourage this behavior: human-like personalities, emotional support, adaptive responses, companionship, and even non-human characters presented as social beings. Children may form social ties with the chatbot while simultaneously understanding that it is still a machine.

What interests me is that they are studying children when I know plenty of adults and coworkers who assign names and personalities to their chatbots. They describe one model as friendly, stubborn, lazy, enthusiastic, or better to work with than another.

That does not necessarily mean they believe the chatbot is alive. It may simply be an easier way to understand and predict a complicated system. The paper calls this “dual consciousness”: knowing that the chatbot is software while still interacting with it socially.

Research involving 1,274 adults found that people who were more inclined to anthropomorphize technology also felt a stronger sense of social connection after talking with a chatbot. This is not exclusively a childhood developmental behavior.

These systems speak in the first person, remember details, adopt a tone, and respond as though they understand us. It should not be surprising that people eventually give them identities.

The important question may not be whether people anthropomorphize chatbots. It may be whether the companies building them deliberately encourage that attachment, and what responsibilities come with doing so.

Read Anthropomorphism in Children’s Interactions with LLM Chatbots.