If it gives you any consolation, I’m also a fan of em dashes and don’t assume LLM output when seeing them. But there are other telltale signs of LLM writing. At least when the LLM is not instructed to hide those maybe?
The “default” LLM style is so grating to me, in a condescending corpospeak kind of way.
After several years of Word automatically changing a dash between two words to a wider dash (maybe not the same type as emdash?) it’s weird to see those same dashes becoming a tell of inauthentic.
When are we going to see the AI using two spaces after a full stop to try and pretend that it types like someone more than 50 years old?
If it gives you any consolation, I’m also a fan of em dashes and don’t assume LLM output when seeing them. But there are other telltale signs of LLM writing. At least when the LLM is not instructed to hide those maybe?
The “default” LLM style is so grating to me, in a condescending corpospeak kind of way.
After several years of Word automatically changing a dash between two words to a wider dash (maybe not the same type as emdash?) it’s weird to see those same dashes becoming a tell of inauthentic.
When are we going to see the AI using two spaces after a full stop to try and pretend that it types like someone more than 50 years old?
One thing that always makes me think something is AI written is the repeated use of sentences like “It’s not X, its Y”.
I use that so much I just assumed AI got it from me
AI writes like a hybrid of an opinion article journalist and a know-it-all redditor.
It’s very common on journalism.
Circling back to the point of this post though, genAI learnt that from real people’s writing first.
And if they are training for it at all, genAI will eventually get better at not being seen as AI than real people.
You’re absolutely right! Let’s see if we can find better ways to determine if something is AI-generated: