I wanted to test out if ai checkers are actually legit, so I made chatgpt generate a prompt and I pasted it into the ai. It said 0% ai so it's written by a human ? then I pasted my African friend's new story inside and omit said 27% ai mix and I was like wtf, she didn't use ai.
So the ai checker is dumb, how can you say 0% ai to something chatgpt literally wrote but 27% ai to what an African lady wrote? I also did some more research and it stated that ai checkers tend to consider work written by non English natives ai work
Maybe I wasn't using the right ai checker
This is a huge problem with AI in regards to written language. It's been proven time and time again that it's biased against non native speakers regardless of language.
The problem is, the general theory to test for AI is to look for patterns. Native language users tend to oscillate and may or may not have set patterns because the language was learned in a natural environment. Non Native language writers and speakers trip up ai checkers detection systems because they tend to fall into patterns.
You know, the patterns you would teach someone in school?
Like when I learned spanish one of the main ways it was taught to me was repetition of common phrases.
Those patterns cause AI to flag, and most of those companies don't really care.