As usual with AI-generated artwork: looks nice at first sight, but if you look closer, you can't help but notice the flaws. E.g. the ambigrams: in the "happy"/"holiday" one, the second word is actually missing the "i", and the two "blessing"s are really hard to read. Also, the "campfire man"'s face seems to be melting in a very disconcerting way...
I'm a photographer, and for years I've been pixel peeping at photos taken on phones with "portrait mode"; many years after the first introduction of the feature, regardless of the implementation, results still look crummy to my eye.
Looking at fine elements like hairs (nevermind curly hair) is a disaster, especially when you're used to fine classic german/japanese optics that accurately reproduce every subtle detail of a subject while having extremely aesthetically pleasing sharpness falloff/bokeh.
I've had to swallow the pill though: No one (end users; pros are another story) cares about those details. People just want something that vaguely looks good in the immediate moment, and then it's on to the next thing.
I suspect it'll remain the same for AI generated visuals; a sharp eye will always be able to tell, but it won't really matter for consumption by the masses (where the money is).
