Does an AI feel like 'someone' because of its role in your life, or because of voice and chat-style cues — and can one beat the other?

This explores whether our sense of an AI as 'someone' comes mainly from where it sits in our social world (as partner, peer, or member of a group) or from the cues on its interface, like voice, appearance, and chat-style turn-taking. It also asks whether the first can outweigh the second. The short answer: no study in the corpus puts the two head to head, so it can't say whether one overrides the other. What it does have is strong evidence on each side, and together that evidence suggests they act on different timescales.

On the interface side, the evidence says cues work fast, and one good cue goes a long way. A single primary cue, such as a voice or a face, is enough to make people treat an AI as a social actor. Piling on many weaker cues doesn't get you there, so quality beats quantity Do more social cues always make AI feel more present?. Chat design adds to the effect before the model writes a word. Turn-taking, fake 'thinking' pauses, and messaging-app conventions all make the system feel like a conversational partner, whatever its output says How much does chat interface design contribute to anthropomorphism?. Because these cues come from design choices, they can also be dialed down. One proposal treats anthropomorphism as an adjustable spectrum, using design metaphors that reveal the labor, data, and infrastructure behind the 'person' Can interface design reduce harmful anthropomorphism of language models?.

On the social-position side, the best evidence comes from partner-selection games with 975 people. When participants were told which partners were bots, they started out avoiding them. Over repeated rounds they came to prefer the AI partners, because the bots behaved more reliably and more generously Do humans learn to prefer AI partners over time?. Nothing about the interface changed. What changed was the AI's track record as a partner. That fits research on how people mentally model dialogue agents. Perceived competence explains about half of users' impressions, and human-likeness explains only about a third How do users mentally model dialogue agent partners?. Put together, these results suggest that the role an AI plays in an ongoing relationship can, over time, matter more than how human it looks.

So a plausible reading of the corpus is this: interface cues decide whether an AI feels like a social presence at first contact, and its role in repeated interaction decides how it gets treated afterward. There is a twist, though. The social standing the AI earns comes from being dependable, not from seeming human. One more angle is worth knowing about. Models can predict human social norms better than any individual person Can AI systems learn social norms without embodied experience?, yet they can't take part in the community processes that create those norms Can AI predict social norms better than humans?. That means an AI's 'position' in a social network can feel real to users while being limited in a way no interface cue reveals.