The Rise Of AI Writing And The Decline Of Human Voice

Artificial intelligence has become a powerful writing assistant, helping people draft emails, essays, marketing copy, and social media posts in seconds. But as these tools grow more popular, researchers are raising concerns about an unintended consequence: AI may be changing not just what we write, but how we communicate altogether, according to Axios.

New research suggests that widespread use of large language models is making language more uniform. A study conducted by University of Southern California found that after the release of ChatGPT, diversity in writing styles declined across several forms of communication, including scientific publications, local journalism, and social media posts. Researchers observed fewer differences in vocabulary choices and sentence patterns, pointing to a growing preference for polished, formulaic language.

Axios writes that the influence appears to extend beyond writing. Researchers at the Max Planck Institute for Human Development analyzed more than 740,000 hours of spoken and written material and found that certain words commonly associated with ChatGPT responses are appearing more frequently in everyday communication. Words like “delve,” “meticulous,” “boast,” and “comprehend” have become increasingly common, suggesting AI-generated language may be shaping human speech habits as well.

Morteza Dehghani, who led the USC research, believes this shift is happening because people are becoming familiar with a specific type of polished communication. “People get used to this idealized, very predictable form of language, and even people who are not using it, in order to have that sense of powerful, influential writing, they start writing more like LLMs,” he told Axios.

Not everyone sees that as progress. Alex Mahadevan of the Poynter Institute for Media Studies argues that AI-generated content often feels empty despite being technically sound. He described it as noticeably “soulless” and “mediocre,” adding, “There’s no art in it.”

For Emily M. Bender, the concern is personal as well as cultural. The University of Washington linguist said she avoids AI-generated writing whenever possible, explaining, “I do my very best not to read any synthetic text.” However, she admitted that identifying it is becoming increasingly difficult: “oftentimes people will send me something and I won’t know.”

That challenge may only grow as AI adoption accelerates. According to a 2025 survey from the Brookings Institution, nearly one-third of small businesses now use AI tools for customer service and outreach, while 16% of individuals report using large language models for communication and social media content.

Bender warns that the pursuit of flawless AI-style writing could create what she calls the “‘LinkedIn average,’” where communication becomes polished but generic. Mahadevan echoed that frustration, saying he misses “good bad writing,” the kind of imperfect but memorable work that reflects real human personality. He admitted that AI’s growing presence has even made him second-guess his own style: “I have been second-guessing myself, thinking, ‘well, sh*t, is someone going to think this was written with AI?’”

At the heart of the debate is a larger question about what writing actually does for people. Bender argues that writing is more than producing clean sentences—it helps people process ideas and sharpen their thinking. “There is value in the struggle of writing, because we learn to express ourselves, and we learn to do the thinking that happens as we’re writing,” she said.

As AI tools become a permanent part of modern communication, experts say the challenge will be maintaining individuality in a world increasingly shaped by machine-generated language. “Each time we choose not to do that, we are losing out, both individually and societally,” Bender says.