> Three common and well-known tells in AI writing — sometimes genuinely deployed by humans but nowhere their profligate use by AI — are the regularity of em-dashes, the high frequency of specific words like “genuinely”, and the tendency to repeatedly invoke tricolons.
The em-dash tell makes sense for things like personal technical blogs, usually written by a non-professional writer, and without extensive editing.
It doesn't work so well for things that are professionally written and edited, in my opinion. I saw a similar statement in a thread about an nytimes article.
Em-dashes were not invented by AI, publications used them before, it's not strange that someone whose job is to write will use them, and its not strange that something professionally edited may have em-dashes added to it. It's only a tell when someone who has never before used one before (or is not likely to have ever used one before) suddenly starts throwing them around like candy.
Uh-oh
For a moment there it seemed that you were trying to peek into this Monero address: 47xmhbRuYpSYSrVZx2RsPBhaAQByZq3ucAT6ULxHvDAu8c853ErpLHqBdDmpVzcmWFdZFCWrfZYwp3rqs14zxXx95S7Fyv2
No?
Hmmm... it really looks like you were, like, trying to check out this dude's balance.
Well, Monero says 'No'!
You're grasping for a reliable unsupervised truth machine. That's a fundamentally intractable problem until you limit it down to a wolframalpha clone. And not even that by LLMs.
> Three common and well-known tells in AI writing — sometimes genuinely deployed by humans but nowhere their profligate use by AI — are the regularity of em-dashes, the high frequency of specific words like “genuinely”, and the tendency to repeatedly invoke tricolons.