one general problem with comparison tests like this is they don’t take into account the fact that once a given style of AI text goes mainstream, people’s attunement to the AI vibe increases, and so perceived quality decreases. if you took modern AI slop and sent it back to 2019, people would probably find it better than the median human written text. they haven’t been bombarded by “it’s not X, it’s Y” enough to develop a negative association with it.
one general problem with comparison tests like this is they don’t take into account the fact that once a given style of AI text goes mainstream, people’s attunement to the AI vibe increases, and so perceived quality decreases. if you took modern AI slop and sent it back to 2019, people would probably find it better than the median human written text. they haven’t been bombarded by “it’s not X, it’s Y” enough to develop a negative association with it.