top | item 46380644

AI Image Generators Default to the Same 12 Photo Styles, Study Finds

15 points| donatzsky | 2 months ago |gizmodo.com

3 comments

order

kadoban|2 months ago

While interesting, the headline naturally overstates the result. They leave out that this occurs if you feed the output back into the input hundreds or thousands of times. Nothing I see here suggests that it means anything for a single generation from a prompt.

soulofmischief|2 months ago

The conclusion is also obvious for anyone even vaguely familiar with attractor dynamics.

metalman|2 months ago

they call AI "visual elevator music"

I call it intermitent wipers for everything.

reality is already 100% saturated, and we are optimised to function and create within the limits imposed.

take sugar, a white crystaine substance that can not be further concentrated, the only way to get a bigger neural bang is to go to another white crystaline substance, with more bang, and so on, till now we have substances where the bang is unsurvivable. this is reality, it applies to visual, auditory, and all of our senses and capabilities, total saturation......now, that does not mean that humans are not gready fucks who want more, more, more, but AI has to give us a greater ability to absorb, more, percieve more, survive more, not just a tired spin on shiny.