The term “junk DNA” is a misnomer. It refers to DNA that does not code for proteins – only about 2% of genes do that – but it unfortunately implies a uselessness of certain DNA. That really isn’t what biologists mean when they use the phrase (or, rather, when others use the phrase; it has been out of vogue amongst professionals for some time now). All they mean is that we have DNA which appears to have no function. This makes sense in the light of evolution since natural selection wouldn’t necessarily be expected to select against useless DNA. After all, why not just leave it there? Unless it constitutes a substantial energy drain, it doesn’t matter. (Note: I am not referencing conserved DNA.)
However, new research is showing that much of our noncoding DNA does serve important functions. Namely, it regulates the genes that do produce proteins. There is still a substantial portion of the genome that appears to have no function, of course. Moreover, there is useless DNA out there that doesn’t code or regulate anything (microsatellites come to mind). However, we’ll all have to wait for further research before we really know the full nature of the human genome.