No not at all. I’m saying all the effort on the “detection problem” is a waste because it’s totally reasonable an LLM and a human would write identical text for a task.

I’m not arguing that humans no longer need to be taught to write well (although I’m unsure about this personally), I’m just saying that trying to test that by detecting if an AI wrote it vs a human is ngmi. Probably gonna have to do in-person exams or oral exams without access to a computer.

Instead of worrying about whether an AI or a human wrote it, we should be figuring out better ways to teach and evaluate critical thinking.