>>38180
A computer solves all problems through computation (almost-but-not-quite calculation). That's why we refer to processing speed in terms of computations per second. Since we're discussing software, it has everything to do with computation. An AI thinks only in terms of computations, so we would have to define whatever we're looking at in terms of computations.
I do agree that it's a body part and nonstandard, which is what I was trying to get at above.
Reiterating for clarity, we draw a lot of conclusions from context; even with a mosaic we can generally tell what the censor hides unless we see only the mosaic.
AIs can't do that specifically because it's nonstandard. Like I said, enhancing an eye to view a reflection is impossible because whatever the reflection is is going to be unpredictable, right? Similarly, decensoring something… You'd need to know the angle it's being viewed at, the scale (That includes ALL scales, from the length and width to sub-parts that may or may not be there, etc) and other muddiness like fluids or disfiguration. Even if it's something that looks fairly simple and not very heavily censored might as well be a stream of random pixels for all a computer cares. A computer doesn't even know what a human looks like, let alone one or more humans in arbitrary poses with a background and possible foreground obstructions, among the countless other things.
tl;dr agree, point was mostly that computers are dumb and don't know what anything looks like