I teach computational approaches to literature at a university, and I always give my students a test to see if they can tell which poems were written by a compu

I teach computational approaches to literature at a university, and I always give my students a test to see if they can tell which poems were written by a computer, and which were written by a human. They tend to score about 50%. But to me, a literary scholar, I know immediately, from a glance even, which ones aren't written by humans, because they're patently ridiculous. I guess if you're not really a reader of poetry, or don't understand it, it's all just nonsense anyway. But for the rest of us, not even the most advanced language models can pull off a poem, convincingly.