News

You’re Not Imagining It. People Actually Are Starting To Talk Like ChatGPT.

I’m both terrified and highly amused.

Devrimb via Getty Images

Freshly delivered among the latest news in the “they’re studying what?” field of academia, the Max Planck Institute for Human Development has released a report that asserts that widespread and frequent usage of Large Language Model AIs, such as ChatGPT, are altering how people speak out loud.

So forgive the em dashes—I was a fan of their limited, judicious use before AI ruined them—while we delve into this intricate realm and underscore how many of these ChatGPT-favorite words I’m adept at working into this story.

Videos by VICE

Who’s mimicking whom?

With the growth of online chat culture in the late 1990s and 2000s, helped along by AOL Instant Messenger (the pinnacle of humankind’s digital communication, honestly), keyboard lingo such as LOL, LMAO, and BRB became standard for everybody.

Some of those lived on into the smartphone texting phase of humanity, others didn’t, but nobody really worked them into their spoken language. To this day I remember the one kid in eighth grade who said “LOL” during class. Poor kid was never socially the same afterward.

As the Max Planck Institute for Human Development’s report suggests, certain words are favored and more heavily used by LLM AIs such as ChatGPT, including (but not only) delve, adept, meticulous, realm, intricate, and underscore.

Overuse of words such as this are hints that a text was created using AI, and these words in particular are some of those the researchers looked at when charted how people’s spoken language has changed with the growth of AI use.

Basically put, the researchers found that the more people use AIs, the more they tend to use the AIs’ favorite words in their own speech.

Is ChatGPT the Problem?

TechRadar’s Eric Hal Schwartz made a good point, though, that other examples of widely popular, technology-driven platforms have worked their way into commonly spoken English, such as saying “hashtag” before a word or phrase to hint at… well, almost anything.

Thank Twitter for that. When somebody would do so, it was almost always tongue-in-cheek and self aware. There was a nod to its own cringiness that meant the one saying it was in on the joke.

Read the entire report if you like. It’s only nine pages, so while it’s dense with the sort of academic-speak you’d expect, it does have a few meticulous gems worth uncovering.