Sidestepping the obviously controversial bit, a big problem with doing this is that UBI is relatively new and not well-understood, so so a program that targets a specific group is bad for both itself and the concept of UBI because it only collects very specific data that are easily dismissed because of selection bias.