i saw an interesting video the other day of some folks with actual background in poetry looking at the examples from that study, and the only way the ai was able to consistently fool them into thinking its output was human written was by plagiarizing passages from real poems verbatim 🤷♀️
anyway i'm harping on about that study in particular because i'm annoyed that it's being presented as good science when its methodology is so flawed: the criteria the authors asked participants to evaluate reflect, well... just a vapid perspective of art, to be frank. is it any wonder that ai excels in those conditions?
=> More informations about this toot | View the thread | More toots from sjolsen@tech.lgbt
text/geminiThis content has been proxied by September (ba2dc).