Some viruses contain RNA rather than DNA. These viruses insert their RNA into a living host cell where it is reverse transcribed into DNA. This virus-encoded DNA may be inserted into the host DNA from where it can be transcribed into more viral RNA and proteins on the way to making more virus particles to release into the environment again. Viruses that work in this way are called retroviruses.
Iin the 1970s, upon finding that cells of chickens contained sequences complimentary to a known retrovirus that infects these hosts, scientists assumed that viral DNA had become resident in the chicken and this information was passed down through the chicken generations as an “endogenous retrovirus.” Thus, DNA sequences, that resemble viral RNA sequences, are considered to constitute ERVs.
There are now claims of shared patterns of ERVs in humans and chimps. These claims depend upon certain assumptions. One is that humans and chimps share a history of common descent. The other assumption is that the “ERV sequences” have no function. If ERVs do exhibit function, then similar patterns in humans and chimps could be an example of common design. Thus these patterns might not have come from retrovirus action at all. It could be that chimp and humans both exhibit a given sequence because it contributes to their health and success rather than their both having inherited a “mistake” in the DNA sequence from a common ancestor.
Many ERVs exhibit important functions such as in embryonic development. There is good evidence that ERVs are a class of DNA that is functional. ERVs are not an argument for “junk DNA” and human evolution.
The bottom line is that we should not allow ourselves to be swept along by arguments based on dubious assumptions. Always find out what the assumptions are before continuing further into the implications of any set of arguments. Casey Luskin for example concludes that even a common pattern of ERVs shown by chimps and humans does not necessarily mean anything: “If ERVs are functional, however, then that means even if they are ‘shared’ and fit into a nested hierarchy, they need not necessarily indicate common descent but could just as easily reflect functional constraints and common design.”