What you need to know is the rate at which insertions were occurring during the time that led up to the split into gorillas and us, and the how long that period was.
The first I can make a rough estimate of. Take a modern human, chimpanzee and gorilla chromosome and trace them back to the point when the gorilla line split from the others. Only insertions that were polymorphic at that point could experience ILS in the three species. Polymorphic insertions would be those that occurred between that point and the most recent common ancestor of those chromosomes. The mean time to the common ancestor for 3 chromosomes is (mumble) 8/3 * Ne, where Ne is the effective population size at that period. The effective population size then is difficult to estimate precisely, but the evidence is that it was pretty large, something like 50,000 maybe. That would give a time to the MRCA of 130,000 generations, or about 2.7 million years.
That number times the rate at which ERVs have entered and fixed in our history will give you the total number of candidates for gorilla-human shared ERVs. Roughly 15% of our genome is most closely related to gorilla, so ~15% of the candidates should be shared only by humans and gorillas (or only be seen in chimp). About 4 times as many of the candidates should follow the species genealogy, and be either only in humans and chimps, or only in gorillas. Assuming the insertion rate was constant (not a good assumption, unfortunately), the number of human-specific insertions -- those that occurred after the human-chimp speciation) should be roughly 2 times higher than the number of ILS candidates (since the human-chimp speciation time is ~2 times 2.7 million years), or about 13 times higher than the ERVs that follow the human-gorilla vs chimp genealogy (2 x 1/.15). How many human-specific ERVs are there?
That's a very crude estimate. It doesn't take into account that ERV insertion rates have varied, and it ignores some other things too. (And of course, I could have done something wrong.) Getting a proper answer would require some real modeling.
ETA: Correction: I was assuming that ERVs that ended up either in the human-gorilla line or in the chimp line (in regions of the genome where humans and gorillas are most closely related) should be counted, but that's wrong. Those ending up in the chimp lineage will look like chimp-specific insertions, and not violate the species tree. So cut my above estimate of tree-violating ERVs by a factor of two; that makes my crude estimate be that the number of human-gorilla shared ERVs (missing in chimp) should be about 4% of the human-specific ERVs.