This just isn't true in my experience. 2026 LLMs, especially if prompted well, are much better at writing than most humans. I think journalists really underestimate how bad most people are at writing.

Here, Kriss correctly identifies this ChatGPT prose as being meaningless garbage. But spend any time on Tumblr and Reddit and you'll know that people write like this *all the time*.

@ShakeelHashim They are excellent at producing grammatically-correct and persuasive writing that is cheap to produce, but increasingly expensive to verify. "But isn't that what a good writer is?" Depends if you want a world where "good writer" can mean a con artist who plagiarizes very well.

@ShakeelHashim Could you share a prompt or chat transcript that produced substantially better writing than most humans by your lights? I could see lots of people being interested in taking a look.

@ShakeelHashim He is categorically opposed to AI writing for reasons that I find reasonable. Maybe most people can't string three sentences together, but spiritually they are better writers than AI

@ShakeelHashim True. But in my experience this pertains to English only, or shorter pieces. Both 800w+ or Italian writing in general still falls short. (5.5 xhigh, O4.6/4.7, 3.1 Pro/3.5 Flash, with or without extensive prompting, reference material, etc.). O4.6 comes closest but no cigar.

@ShakeelHashim Yes, structurally much better. Not (yet) better at picking interesting ideas and expressing them, though useful refinement tools here as well. IMO it works extremely well if you bring your own ideas and give it some samples of your best writings then iterate.