The Linguist Arguing That AI Chatbots Could Reshape How We Talk to Each Other

Maryellen MacDonald, the Donald P. Hayes Professor Emerit in the Department of Psychology and the Program in Language Sciences at the University of Wisconsin-Madison, has built her research around a claim that runs against intuition: producing language, not just consuming it, is what sharpens the mind. Now she is applying that framework to a practical question, whether the way people speak to AI systems bleeds into how they speak to one another.

MacDonald studies how people produce, learn, and comprehend language, according to her contributor profile at Psychology Today, where she writes a blog titled "The Power of Words." Her work has examined how the act of producing language has downstream effects on the brain, how working memory connects to language ability, and how people unconsciously absorb the statistical patterns in their language that make comprehension and production easier. She is the author of a general-audience book, More Than Words: How Talking Sharpens the Mind and Shapes Our World.

Why the argument matters now

The distinctive part of MacDonald's position is where she puts the emphasis. Much of the popular conversation about brains and language centers on reading, vocabulary, and comprehension, all forms of taking language in. MacDonald's research points in the opposite direction, treating speech production as a form of cognitive exercise. In one of her recent posts she frames talking as a workout for the brain, and raises a concern that opportunities to talk are shrinking as attention shifts toward screens.

That concern connects directly to the AI question. In a post titled "Will AI Make Us More Rude?", MacDonald raises the possibility that the tone people adopt when addressing chatbots, often clipped, transactional, and free of social friction, can carry over into human interaction and wear down social norms. It is a hypothesis about behavioral spillover rather than a claim about the technology itself, and it sits closer to observation than to measured effect.

What is established versus what is proposed

MacDonald's academic standing rests on decades of published work on language production and working memory. The applied claims in her public writing, that phone use is reducing how much people talk and that AI conversation styles may reshape human ones, are framed as concerns and areas of expert worry rather than settled findings. The distinction matters for readers trying to separate peer-reviewed cognitive science from forward-looking commentary about emerging technology.

For a technology audience, the value in MacDonald's framing is that it reroutes the debate about AI and cognition. Most attention goes to whether AI degrades reasoning or memory through overreliance. Her work suggests a quieter mechanism worth watching: the everyday texture of how people phrase requests, and whether interacting with systems that never take offense trains people to expect the same from each other.