Re:Glaze (Poisoning AI Artists)
by michaelmrose on Sunday March 19, 2023 @11:58PM
(#35227108)
Why would it need to specifically circumvent this instead of casually circumventing it simply by better mapping how the bits on a program map to how an image appears to the viewer?
if f(x)1011 and 1010 both "look" the same shouldn't a well designed AI learn equally well to recreate that particular look. It doesn't need a "deglaze" circumvention step it just needs a continually improving learning process that automatically accommodates both glazed and un-glazed images.
Re:Glaze (Poisoning AI Artists)
by kevingadd on Monday March 20, 2023 @01:12AM
(#35227544)
Right, a tool explicitly designed and used to "deglaze" images would be different from a network that just happens to not be affected by glaze, I expect.
if f(x)1011 and 1010 both "look" the same shouldn't a well designed AI learn equally well to recreate that particular look. It doesn't need a "deglaze" circumvention step it just needs a continually improving learning process that automatically accommodates both glazed and un-glazed images.