Yes, I don’t think it’s a matter of training.

The diffusion model generates pictures by starting on a canvas with random pixels, then it edits those pixel colours and carves the picture out of that chaos

To achieve an area with all the same colour, it would need to put very exact values on the last generation step.

It can be fixed easily with a very subtle lowpass filter, but that would be human intervention. The model itself will have a hard time replicating it

Two that I noticed are:

For drawings in the ghibli style, you can see noise on areas that should have all the same colour. That’s because of how the diffusion model works, it’s very hard for it to replicate lack of variation in colours. If fact that noise will always exist, it’s just more noticeable on simple styles.

For music, specifically with Suno, it tends to use the similar sounding instruments between different tracks of the same specifispecified genres, and those sounds might change during the track and never come back to their original sound (because it generates section by section of the track from start to end, the transformer model will feed the last sections back as input to generate the new ones, amplifying possible biases in the model)

It is conceivable.

For example, imagine a society like ours but where everyone, no matter their wealth, has to do essential jobs, taking turns. For example everyone in your city needs to be a garbage collector for one week every 2 years, or they need to work in hospitals to help clean patients for a week every year, maybe you need to work in the fields for a month every 2 years, basically all jobs that people only do because they can’t do anything less tiresome plus jobs that are now almost fully automated to produce essentials but still require some labour.

In that system, you’d always have enough workforce to give everyone enough food, homes, healthcare and education to live, while people might still work at secondary non-essential jobs, voluntarily, and gaining a bit more to have their fancy cars and yatch.

This is a conceptual society where, despite the possibility for individual differences, you don’t really have classes, because no matter if you were born in poverty or you are Elon Musk, you all have to take part on essential services equally.

I recently made a pdf with some of my notes on hints of AI in images and music, but I’m not sure how to send files here

It’s not easy ofc, and it will get harder with time, but I am convinced we can tell if trained a bit. Because there are clear differences in the creative processes between humans and machines, which will always result in different biases

With time I think we’ll learn to only trust people we have some social connection with, so we know they are real and they don’t use AI (or they use it up to a level acceptable to us)

This could go wrong in so many ways