Did you notice at 5:48 the narrator asks Spot if it "could show us its parents?" whereupon Spot went over to the exhibit of Spot V1. It was not programmed to do that! Did it "know" that it was an earlier version of itself? I don't think so but probably encoded in its training data a connection was established between parent, progenitor, previous version, prototype, and its visual memory.
Although we poopoo a lot of this type of behavior in AI don't we all often react to questions without what we term as "thinking"? In this case we usually have the advantage of being told who our parents are.