Comment by yosefk
12 days ago
I don't love the political agendas behind many of the attempts at AI safety, but it's not "just like humans." Humans understand what they shouldn't say; "AI" gives you black Nazi images if you ask it for "diverse characters" in the output which no human would do. A big theme in all of these things is that AI isn't and thus all attempts to make it do this or that have strange side effects
> which no human would do
Give someone not familiar with history the same task and they'll do exactly the same.
Or actually, give someone familiar with history the same task and yell at them every time they don't deliver diverse characters, and eventually they'll learn that you consider diversity more important than accuracy or context, and do exactly the same.
The fact that it gives you these things means that humans would do it, because the training data includes exactly these things.
I'm fairly confident there's virtually no ethnically diverse nazis in diffusion models' training set.
It simply has a model of what ethnically diverse people look like, what nazi uniforms look like, and combined the two when asked.
The training data includes imagery that, when interpolated over a high dimensional manifold, results in these things.
That doesn't imply that they were in the training set, or even anything close to them.