How Talking to AI May Be Reshaping the Way We Talk to Each Other

The habits we form when barking commands at chatbots may not stay contained to our screens. In a series of columns for Psychology Today, Maryellen MacDonald, a Professor Emerit of psychology at the University of Wisconsin-Madison, argues that the way people speak to AI systems can spill over into how they speak to one another, potentially fraying the social norms that hold conversation together.

The concern is straightforward. When a user types a terse instruction to a chatbot, there is no penalty for skipping the softeners that mark ordinary human exchange: no please, no thank you, no reading of the other person's reaction. MacDonald's argument, laid out in a July 2026 column titled "Will AI Make Us More Rude?", is that repeated practice with this stripped-down mode of communication can carry over into interactions with actual people.

The brain-as-muscle framing

MacDonald's broader thesis is that conversation is not just a way to exchange information. It is cognitive work. In a June 2026 piece she called "The Word Workout," she frames talking as a form of exercise for the brain, and she is the author of a book, "More Than Words: How Talking Sharpens the Mind and Shapes Our World."

The implication runs in two directions. If regular conversation strengthens cognition, then a decline in talking, as attention shifts toward phones and screens, could represent a loss. MacDonald writes that experts are concerned about exactly this decline. And if the conversational practice we do get is increasingly with machines that require no politeness and offer no social feedback, the quality of that practice changes too.

Why this matters beyond the individual

The framing moves the AI-and-society debate away from the usual questions about misinformation, jobs, or model accuracy, and toward something more granular: the micro-behaviors of everyday speech. Social norms around courtesy are learned and reinforced through repetition. If a growing share of a person's daily verbal interactions happens with a system that treats bluntness as efficient, the argument goes, the reinforcement loop weakens.

It is worth being clear about the limits here. MacDonald's columns present a hypothesis grounded in her research on language and cognition, not a measured finding that AI use has made a population ruder. The claim is a warning about a plausible mechanism, not a documented effect with an attached statistic. No study cited in the columns quantifies how much command-style AI use changes human-to-human politeness, or over what timeframe.

The Asia context

The question carries particular weight across Asia-Pacific, where AI chatbots and voice assistants are being folded into daily life at scale and where several languages encode politeness directly into grammar. Japanese and Korean, among others, build honorifics and formality levels into verb forms and pronouns, meaning the choice to be polite is not optional decoration but structural. How those systems interact with machines that default to flat, register-neutral responses is an open question that the English-language framing does not fully capture.

For developers building conversational products in the region, the practical takeaway is less about grand social decline and more about design. The tone a system models, and the tone it rewards, is a choice. Whether that choice nudges users toward or away from the conversational habits they carry into human relationships is, for now, an untested proposition worth watching rather than a settled one.