Maybe I am a bit of a curmudgeon on this subject. One of the biggest flaws I see in this is the pervasive "mirror imaging" of the characters. Its simply NOT dynamic and not believable. The objective should be to emulate "real life" to the point where it is hard to distinguish between AI and real life action. Real people do not ever appear to be of identical shape and size, standing or laying prone in directly opposite directions. Since I have no experience using AI and really am not very interested in doing so, I have to presume this is either a consistent limitation across platforms, or that the users are "all" taking a short-cut by making the opposing parties one size and shape. Until the product output becomes truly dynamic and 3-dimensional, I don't think I will be impressed with the technology.