The idea of junk DNA is based on the fact that it doesn't code for any proteins, but many other functions have been found for it (small nuclear RNA, microRNA, small interfering RNA, etc.). Some parts of the genome are not transcribed into anything but still have a functional purpose, such as in telomere caps and in folding. And there are large parts with no known purpose, they might be remnants of working genes, and they might have a function in evolution (see the "might"). One research project (Encode) found that around 80% of the human genome is transcribed, but the argument against this is that DNA being transcribed may not necessarily mean it has a function. The theory of junk DNA hasn't diseapered but it isn't necessarily true either.

I am just a student, so take my info with a pinch of salt.

For real, I thought he was going there. Like : the AI keeps trying to fight off the coders messing with its perfect code so it keeps generating junk code to protect the actual code.