If you think throwing life into a black box and consuming its results will improve productivity and enhance creativity. You're forgetting the only thing we feed this thing is 'results'. You're drinking the piss from a million artists and thinking it's fresh water. Eventually this filtered, censored source of creativity is creating for AI and not humans because humans quit being a source of material and nothing more than a requester.
The short sightedness of believing you can commodify the human spirit is the most incredulous shit I've ever seen.
And seemingly harder to reach non-monopoly/oligopoly solutions.
Paying people to move goalposts is expensive, and it might turn out to be a flop, but you could do both with Sora cheaply, and even make it into an interactive movie which, something impossible with conventional read only films.
They skirt around the important details, in that in all those analogies the entire process/field wasn't replaced/impacted by automation, only facets. Advanced generative NNs aren't merely dumb automated tools that still require human craftmanship but rather, at a baseline, take the place of the humans in virtually all respects of design, implementation and continually synthesizing based on others' work.
The analogy presented in your post doesn't account for the designers still involved with the process, the real-world practicalities that involve humans of getting clothing made even with all the automation available. Generative AI doesn't have this goal; it's increasingly allowing for replacement of all humans and processes involved, besides the one wishing upon a genie lamp.
Now one can of course apply their own craftsmanship to such outputs if desired and there are examples of this from existing artists/writers where it fits more like a tool into a workflow (and such analogies would actually apply here), however as we can already observe the vast majority of uses don't and there is still at a minimum a usable result, as the tech advances.
There are also parallels one commonly reads drawn to how humans operate: if a human studied others' artwork then they too would be able to synthesize results. However no human has memory and mind's eye recall like that of advanced NNs, which is plainly observable from the output of human history. It's an unfair comparison. The goal of generative models is to be superhuman yet only human arguments are being applied.
I think there are a bunch of positives and negatives to this but it's hard not to see the impact it will have particularly on human motivation and societal expectations, regardless of financial incentives, apart from the general shut-up-and-take-it attitude one sees. There are some pretty impactful legitimate concerns to be had that warrant thoughtful discussion.
This is exactly why we need to BAN weaving machines. The will to live is sustained by the work of the skilled artisans who create clothing by hand with looms.
"But I don't want to pay 500 dollars for a t-shirt", you say. What astounding arrogance! Stop destroying culture and accept that we need to protect this valuable craft from technology that threatens to steamroll it.