There will always be a bias towards some default settings in whatever popular tool, because people who aren't very discerning or dedicated will just punch in a few common ideas into the tool and take the first thing that looks good enough to them.
But people on the outside will continue to have an instinctive sense for these default style biases and be able to tell pretty well that an image was produced that way.
The escape hatch will be what amounts to AI stock photo industry, where practiced "AI Art" designers prepare images that are more subtle and unique and sell them pretty cheaply.
The technology surely can be pushed to the point where a well-crafted image is virtually indistinguishable, but most people are going to just drop "anxious woman at keyboard" or "happy dog with a ball" and there's only so much unique information contained in those prompts so defaults will always reign.
> There will always be a bias towards some default settings in whatever popular tool, because people who aren't very discerning or dedicated will just punch in a few common ideas into the tool and take the first thing that looks good enough to them.
How is that different from the established practice of using stock photos?
And most people see stock photos on a blog and see that a bit negatively too. If the ceiling here is stock photos, that's not great, but the issue is that the state of the art is "stock photo but with glaring weirdness"
There will always be a bias towards some default settings in whatever popular tool, because people who aren't very discerning or dedicated will just punch in a few common ideas into the tool and take the first thing that looks good enough to them.
But people on the outside will continue to have an instinctive sense for these default style biases and be able to tell pretty well that an image was produced that way.
The escape hatch will be what amounts to AI stock photo industry, where practiced "AI Art" designers prepare images that are more subtle and unique and sell them pretty cheaply.
The technology surely can be pushed to the point where a well-crafted image is virtually indistinguishable, but most people are going to just drop "anxious woman at keyboard" or "happy dog with a ball" and there's only so much unique information contained in those prompts so defaults will always reign.