Yeah. Very cool. And I love thinking about the objects he generates as perceptual objects for some potential alien being.
But the really cool thing will be when we implement this for a system which works like human vision!
Something which generates “objects” that correspond to objects that people see. Generating them. Novel objects. The way the objects Domas generates are novel, but still meaningful.
Something that doesn’t depend on trying to label every last speck of dust on the planet so it can be “learned” beforehand, the way we do now.
I see what Tesla is doing, heaps of money thrown at artificial situation generators, to generate beforehand every conceivable thing that might happen, so every conceivable thing can be “learned” before it happens.
Or language transformers, only novel by interpolation over stuff that’s been said before.
So much money and effort being wasted.
