Did people stop being polite to computers simply because computers stopped feeling new?

This explores whether the instinct to treat machines as social beings belongs to a technology category for good, or whether a whole category can lose it as it becomes ordinary. The most direct evidence says it can. A 2022 direct replication of a classic study from the 'computers are social actors' (CASA) tradition found that people no longer showed politeness toward desktop computers, an effect that had once been a textbook finding Does CASA theory still shape how people treat computers?. That means the original result may have been capturing a moment in history, when computers were strange and new, and not a permanent fact about how human minds work.

The same fading shows up at smaller scales too, and they line up with the category-level story. In long-term studies with the chatbot Mitsuku, the social processes that pull people into a relationship weakened predictably as interactions repeated, so single-session studies can't tell you how a chatbot relationship will look months later Do chatbot relationships lose their appeal as novelty wears off?. Persuasion follows a similar curve: Claude and DeepSeek started out more persuasive than human persuaders, but that edge wore away over repeated rounds while the humans held steady Does AI persuasiveness fade across repeated conversations with the same person?. With people, rapport usually grows over time. With AI, at least here, the social pull seems front-loaded. If individual users get used to a system this way, it's plausible that a whole society does the same thing to a whole category, only more slowly.

There's a catch in how we explain this. The usual CASA account says people respond socially to computers 'mindlessly', on autopilot. But Lee points out that most studies work out mindlessness backward from the social behavior itself, so the explanation can't be proven wrong Can mindlessness in CASA effects be measured independently?. If politeness toward desktops has disappeared, we can't cleanly say whether people became more mindful, whether desktops stopped giving off social signals, or whether the old effect was never as solid as we thought. Aging out is real in the data, but the reason behind it is still open.

That leads to a useful twist: a category may not age out evenly. Research on social cues finds that one strong primary cue, like a voice or a face, is enough to make an AI feel present as a social actor, while piling on many weaker cues isn't Do more social cues always make AI feel more present?. A beige desktop has almost no primary cues, so once the novelty faded there was little left holding up the social response. Voice-driven, conversational AI may hold on to it far longer, or keep bringing it back with each new feature. Memory matters too: personalization built on what a system remembers from earlier conversations made people share more and find it less creepy than personalization built from intake surveys Does personalization type shape how people feel about sharing?. That suggests designers can work against habituation.

The corpus doesn't yet include a long-term, category-wide study of LLM chatbots, so we can't say whether today's AI assistants will go the way of the desktop. What it does show is that 'people treat X as social' should be read as a finding with a date attached. And the more interesting question may not be whether the response fades, but which design features, like voice, memory, or a strong persona, keep resetting it.