If you look closely at AI and how it's taking shape, you'll notice how it's already developing a standardized form of groupthink. And I don't just mean images. The words might sound perfect and the images might appear crisp but the actual freedom to be creative is stunted.

I can't judge that for the AI applications that already exist. I just read in various places that most publicly available AI is still too dumb to impress. We'll see where the development goes. It is likely to improve and it amounts to much of what is currently produced by humans (especially online, text and image creation or research) will be done by AI in the future. I don't approve of that, but I don't see how it can be prevented. (Though I often use deepl.com which is an AI driven translation app, and it's excellent).
Except that humans themselves create extraordinary works and reach or get the attention of a human audience. Which I'm sure will just be the famous exceptions again.

I've experimented myself, running some old writing through the machine, and the first thing it did was remove all the jokes.

LOL, how humourless. Paradoxes are another thing that an AI certainly can't "grasp" as they are probably very difficult (if not impossible) to program, much like humour.