tbh I'd expect the opposite - inherent bias / regression to the current norm comes through quite consistently when they're used to summarize stuff, and that'll badly taint science using it.