The problems in AI writing, IMO are due to the following factors:

1. Information density. LLM love to shovel in so much packed information in a paragraph.

2. Stiffness in writing style, especially influences from academic writing. Things like "it is not A, it is B" is clearly stemming from dialectic. (man I wish LLM can use more dialectics like syllogism)

3. Following the roots in dialectic, the way that "chat bots", or dare I say, Socratic dialogue, required both the speaker and the responder to have a very high understanding of the topic to be effective. Yet most people are just plain simple, let alone being subject experts, so both the LLM prompter generates slop， and the reader only reads slop

4. The great ADHDification of the society due to tech-addiction. The rise of TikTok or short-form clips, gives an average human a great gratification in dopamine in the terms of fast and infinite visual of information at your fingertip. The attention span of most human now was dragged down from the mean few years ago, and especially apparent in the new generations of students and children. Many people just go for tl;Dr and want to watch clips or play games after, because reading is deemed painful.

While ironically I do have clinical ADHD Type-I since childhood, and I do used to binge play and binge programming (since I started writing game cheats and botnets, boy if I had a goal I will not stop unless my body stopped me), but I'm completely immune to social media (except during the pandemic when it is necessary for communication).

But again, I think those 4 factors are what causing distrust in this new modern version of producer-consumer problem: the fact that your dopamine receptors can't keep up