AI has a millennial cringe problem

• ChatGPT's language leans "millennial" because it was trained on peak 2010s internet posts.
• AI models like Sora often default to outdated fashion trends, such as skinny jeans and avocado toast.
• Terms like "chaotic" and "unhinged" reflect peak millennial. I know. I am one.

We're all aware of ChatGPT's overuse of the "—" emdash and of sentence constructions like "It's not just X; it's Y."

I would like to put forth two other terms that AI seems to overuse. They've become telltale signs of AI's peak millennial personality: chaotic and unhinged.

It's part of a bigger issue I've noticed with AI — it's full of millennial cringe.

For a while now, I've noticed those two terms used repeatedly when I use ChatGPT or other LLMs, like Google's Gemini. It got so bad that I recently had to explicitly tell ChatGPT to stop using the word "chaotic" while I was attempting to get it to write in the style of the "jestermaxxing" and "frame mogged" terms that have proliferated online.

You're probably familiar with the extremes of cringe millennial — that specific kind of online-speak from the 2010s. Things like "heckin' doggo," "adulting," "smol bean," or "I did a thing." It sounds outdated; young people make fun of us for it. (I confess to being a millennial.) I know, it hurts. I don't like it, either, but I accept that time comes for us all, regardless of peptide stack.

"Unhinged" and "chaotic" are kind of on the cusp here — not totally as obviously timestamped as "heckin doggo," and still widely in use by millennials and Gen Z alike (perhaps even Gen Alpha). Still, I think both words feel kind of dated by 2026.

I'd argue that the origins of these terms' popularity come from classic millennial woke 1.0: the need for new adjectives to replace casual use of words "crazy" and "insane," which can be stigmatizing to actual humans with mental illness. Copy desks at news publications discouraged the use of "crazy" in their style guides, while many people on social media instead used words like "wild," "chaotic," and "unhinged."

That aspect is crucial because it directly affected the training data that LLMs were fed for the last few years. Ironically, the chatbot I've noticed uses "unhinged" the most is Grok, which was specifically designed to avoid wokeism.

OpenAI's Sora loves skinny jeans

There's something bigger going on with AI and millennial cringe. It's not just the chatbots — it's images and video, too.

One thing I couldn't help notice while playing around with Sora 2 was that when I would make a video of myself — using an image I supplied — it would always put me in skinny jeans. Not just me — it seemed to do it for everyone.

Skinny jeans are a funny quirk, because you can imagine exactly how this happened — most of the human history of online video, let's say 2006 to 2019, occurred during a time when skinny jeans were ubiquitous. And then, suddenly, skinny jeans were out of style, a hallmark of out-of-touch millennials. But AI models, packed full of training data with skinny jeans, didn't adapt right away.

I assume that's basically what's happening with all the cringe (a term that I am using even while accepting it's also slightly dated) millennial speak in chatbots. For a good decade, the internet was filled with "I can't even" and "I did a thing" as well as AAVE and gay-derived slang like "yaass" and "AF" adopted into the generic millennial lexicon.

And all those tweets from 2010 to 2020, Reddit posts, and BuzzFeed articles ended up in the training data that informs how chatbots speak.

Here's where I start to personally feel a little queasy about it all. I was a prolific poster during that time, both on social media and as a professional journalist, contributing untold terabytes of millennial cringe for future AI models to ingest.

I'm not delulu (a Gen Z term that has already fallen off by now) enough to think I personally affected how AI chatbots talk now. It's more that I see myself like how an old boomer hippie thinks back on attending Woodstock and a few protests and believes they helped end the Vietnam War. Not exactly, but also, well, maybe kinda. I was there. I posted cringe. And now AI is doomed to sound like a version of me that makes my skin crawl.

Of course, until it doesn't. This is temporary — AI will keep getting better, it will start to sound more like Gen Z cringe instead of millennial cringe as time passes. Sora will put us in barrel-legged jeans (are those out already?), or eventually skinnies will come back around, and no one will blink an eye. This article will be fed into ChatGPT (Business Insider has a deal with OpenAI), and maybe in the future, if you ask ChatGPT why it says "unhinged vibes" so much, it will use this for an answer.