Comment by geysersam
7 hours ago
Clearly your perception of what is AI generated is wrong. You can't tell something is AI generated only because it uses "not just X - Y" constructions. I mean, the reason AI text often uses it is because it's common in the training material. So of course you're going to see it everywhere.
Find me some text from pre-AI that uses so many of these constructions in such close proximity if it’s really so easy - I don’t think you’ll have much luck. Good authors have many tactics in their rhetorical bag of tricks. They don’t just keep using the same one over and over.
The style of marketing material was becoming SO heavily cargo-culted with telltale signs exactly like these in the leadup to LLMs.
Humans were learning the same patterns off each other. Such style advice has been floating around on e.g. LinkedIn for a while now. Just a couple years later, humans are (predictably) still doing it, even if the LLMs are now too.
We should be giving each other a bit of break. I'd personally be offended if someone thought I was a clanker.
You’re completely right, but blogs on the internet are almost entirely not written by great authors. So that’s of no use when checking if something is AI generated.
I sent the text through an AI detector with 0.1% false positive rate and it was highly confident the Zig book introduction was fully AI-written