The Biggest Tell That Something Was Written by AI

I used to read every bit of text I see, e.g. on Instagram captions or comments. LLM-generated text has allowed me to drop this habit, because my brain now sees all the "It's not X, it's Y" and assume AI and stop reading...

I'm still addicted to the candy that is Instagram reels, lots of them now have descriptions that basically summarize what is being played (as if people are blind), instead of adding details, out there someone's farming TikTok for content and reposting them on Instagram. And so the attention economy keeps on chugging.

Gift link: https://www.theatlantic.com/technology/2026/05/how-to-tell-a...

"Recently I asked ChatGPT Pro—which I paid for to summarize financial spreadsheets; I’m not a hater—whether it was capable of viciously dismantling an argument I’d read about the stock market"

Huh, parataxis-embedded-in-em-dash-separated-subordinate-clause, somehow I haven't seen that one before.

It's an interesting article, though the author does the thing that professionals tend to do and project just that wince of extra significance to their particular line of work, in this case writing(programmers and engineers do this constantly frankly). I'm personally a bit more bleak than the author, if so much writing has been replaced by AI cliches which don't add much substance to the message being communicated, is it possible that we never were really communicating that well to begin with? I'm just not entirely convinced that where AI replaces human writing there was much of the transcendent human value the article broadly gestures at.

I will absolutely agree with the article on human hangups being part of the process of discovery. I think its important to be stuck and to be stuck often and restart and do the difficult work yourself, in writing and in other endeavors too.

Just look for the em dash. AI loves them. No amount of pleading with it not to use them will help. Nobody knows how to search / replace them. Well, not your regular non-cretinous mentally stable individual anyway,

This part - "AI programs’ efficiency in generating smooth, grammatical text is irresistible, whether you need a savvy sentence in a job application or a line of banter on a dating app" - made me wonder: who finds it irresistible to surrender their individuality when bantering on a dating app to a sequence modelling algorithm?

The article talks about the writing "elites," who normally are expected to take pride in writing. But what about those who just find it fun and exhilarating? Pride after all is a sin, a predictor famously of decline. And perhaps that's what explains the willingness of those (maybe even including Fairbanks, this article's author) who see their relationship to their writing and their status in that world as one of pride to acquiesce in this moment of literary seppuku.

If you're writing, creating, communicating, to compete and be fêted, then you're predisposed to be seduced by a tool that can so quickly put badges on raccoons. If you're doing it to discover your sense of self, your use of that tool is entirely different and you'll naturally keep its output at arm's length from what you actually own as "your writing."

If things right now are hard to tell, it will be easier in the future to tell looking back. So, in the future lets say 2030 we can look back at the things people write in 2026 and that they swear are "my own words" and that fool almost all of us and see clearly that they were not actually their own words after all.

It's going to be like when we all used cat meme pictures in our slide decks - in the future it's going to be obvious and cringe and a liability now.

It's like postponing a major credibility and trust failure event to yourself by a few years in the future for some feel good attention now.

Uh.. I mean can someone tell me these days when it's not written by AI.

"But here's the thing..."