Yeah, that's sad.

And he's correct. AI "detectors" as he describes are both wrong, useless, and doomed to fail.

But there are better ways to get hints. Specific models tend to overuse certain tokens: they'll pick the same character names for a story, or overuse certain phrases, and if you play when them long enough you start to recognize "oh, that's an OpenAI" model, or "this is dry like Nvidia Nemotron," or whatever.

See EQBench's slop profiles: https://eqbench.com/

But, ultimately, the way to fix this is to go back to trusted sources and citations.

This man is a fucking hero. He’s got the concept built just right; I’ve been trying to slap it together much more awkwardly, for years.

I’m not a robot, I’m just following the rules of the language I learned.

Fascinating stuff, and it makes sense that a more rigid, rules-based version of English would be both taught in ESL schools and learnt by an LLM.

If I start a sentence, "The cat sat on the...",

Me: "mat"

your brain, and the AI, will predict the word "floor."

Oh. I'm not sure that's the best example, as that particular phrase is something many children encounter when learning to read.

Also, British English would close the quotes before the full stop (aka period) finishes the sentence - is that something he's picked up since school or is that something that's changed since the Kenyan model of perfect English was set?

What's stupid about people saying only LLMs use emdash and the sentence lists of threes, is that's exactly how we learned to write throughout our Canadian highshool curriculum.

And if you read dialog or narratation in books, its full of emdash asides in sentences.

I have not known Kenyans to be huge sycophants