Hacker News new | ask | show | jobs
by thinkingemote 18 days ago
"Do you people all have brain injuries? AI writing is almost comically easy to detect"

Some can't and that's fine. I also find it comically easy to detect, but it's a kind of pattern recognition, a skill that takes a bit of investment of time and energy on some internal disposition. It also operates on the unconscious: if it feels off it might well be off. Like map reading, or like listening to jazz, some people just don't can't seem to do it and that's fine. Most people around me can't read code and thats fine.

Also: some others don't care about what they see or how they write and levels of literacy are also lowering, and some others are enthusiastic users of the new technology so have to protect their investment.

1 comments

My way of detecting it has become less about patterns and cliches and more about the content of the text itself. The article gives many examples of AI generated restaurant descriptions and while they are written very similarly, the bigger problem is that the text has no actual meaning in the first place. It's just pointless descriptions of what a restaurant is, but written in different cliched ways. A human could rewrite the same content to not use cliches and it would still clearly be AI due to how worthless it is.