Comment by wkirby
I agree with the sentiment elsewhere in this thread that this represents a "hideous theft machine", but I think even if we discard that, this is still bad.
It's very clear that generative has abandoned the idea of creative; image production that just replicates the training data only serves to further flatten our idea of what the world should look like.
Right, the focus is on IP theft, and that’s part of it, but let’s set that aside.
How useful is an image generator that, when asked to generate an image of an archaeologist in a hat, gives you Harrison Ford every time?
Clearly that’s not what we want from tools like this, even just as tools.