May 2, 2026 -Politics & Policy

AI is changing how we write and speak

Josephine Walker

email (opens in new window)
sms (opens in new window)
facebook (opens in new window)

twitter (opens in new window)

linkedin (opens in new window)

bluesky (opens in new window)

Illustration of a framed painting of a robotic William Shakespeare. — Illustration: Lindsey Bailey/Axios

While AI was trained to write like humans, these large language models are eroding the unpredictability of our writing and shifting the way we talk.

The big picture: Researchers found that AI pushes users toward a more standardized speaking and writing style, reducing variations in sentence structure and vocabulary.

The University of Southern California study analyzed scientific journals, local news articles and social media and found writing style diversity dropped sharply after ChatGPT's release.
ChatGPT's favored words — such as "delve," "meticulous," "boast" and "comprehend" — are showing up more in everyday conversation, researchers at the Max Planck Institute for Human Development found after reviewing 740,249 hours of content.

The intrigue: "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," Morteza Dehghani, a USC professor who oversaw the study, tells Axios.

Alex Mahadevan, chief AI instructor at the Poynter Institute for Media Studies, tells Axios that AI writing is noticeably "soulless" and "mediocre," even though it is grammatically correct. "There's no art in it."

Between the lines: Emily Bender, a linguist at the University of Washington, said she does her "very best not to read any synthetic text." But "oftentimes people will send me something and I won't know."

AI use is on the rise, making it difficult for those seeking to avoid LLM-generated content. According to a 2025 Brookings survey, 32% of small businesses use AI for customer service and outreach, and 16% of individuals said they use LLMs for communication or social media.

Zoom out: Bender warns that chasing "ChatGPT level polish" dulls authentic voices and encourages what she likes to call the "'LinkedIn average," or the bland corporate speech users on the networking site often use.

Mahadevan says he misses "good bad writing," in which prose is so poorly written that it's engaging and uniquely human. He added that he now tries to avoid the supposed telltale signs of AI in his writing, such as em dashes.
"I have been second-guessing myself, thinking, 'Well, (expletive), is someone going to think this was written with AI?'"

Zoom in: "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," Bender says.