When you get used to talking with an AI, do you develop a whole new way of relating to it — separate from how you talk to people?

This explores whether people who use AI a lot swap their human social habits for new, machine-specific ones, or end up carrying both. The clearest answer in the corpus is that they keep both. Extended research in the 'computers are social actors' tradition finds that people don't just reuse the scripts they use with other humans. Through repeated use they build a second set of scripts tailored to media agents, and they apply these just as automatically as the human ones Do humans apply human-human scripts to AI interactions?. Studies that track people over time show their responses change in consistent ways as AI becomes familiar. The new script grows next to the human one and doesn't overwrite it.

One way to see why both survive is how people size up a dialogue partner. When users rate a conversational agent, their impressions split into three factors: competence (by far the biggest), human-likeness, and how flexibly it communicates How do users mentally model dialogue agent partners?. Competence is a test you'd give a tool, and human-likeness is a test you'd give a person. People run both at once, which fits a world where machine scripts and human scripts sit together, each covering a different side of the same exchange.

The corpus also suggests the human script may never fully retire, because the AI keeps calling it up. One argument holds that AI output carries the surface signs of real speech, picked up from training data, without the actual back-and-forth that makes speech a real exchange. Users fill that gap themselves and turn the text into something that feels like a conversation Does AI generate genuine utterances or just text patterns?. Shanahan's framing points the same way: a dialogue agent is best understood as a character being played, so treating it like a person makes sense for the character but not for the system underneath Should we treat dialogue agents as role-playing characters?. Seen this way, the AI-specific script may partly be a learned habit of knowing when to drop the human framing.

Designers are now trying to shape which script gets triggered. One proposal treats anthropomorphism as a dial rather than a default, with settings that range from deliberately un-humanlike to fully humanlike, so users are pushed away from trusting the system without question Can interface design reduce harmful anthropomorphism of language models?. Interface critics go further and argue that chat itself is the wrong default for many tasks. It pulls people into conversational habits when clicking and direct manipulation would be faster Is chat the wrong default interface for AI products?. Task-focused interface patterns that avoid typed prompts are one way out Can task-oriented UI patterns reduce the friction of AI chat interfaces?. The less obvious point is that the interface may decide which script you use more than your own experience does. A chat window invites human-style conversation, and a canvas invites the habits you use with tools.

A gap worth flagging: the corpus establishes that the two scripts coexist, but it doesn't say much about how people switch between them mid-conversation or what happens when they clash. That's an open edge to keep in mind rather than a settled finding.