Would an AI Have Any Interest in Using VR?
A Puzzling Hypothetical
I posted a stupid ‘humor’ piece recently in which I tried to illustrate some of the similarities and differences between artificial intelligence and virtual reality. As I neared the end of the post I started wondering what an AI would actually make of virtual reality. Specifically what an embodied artificial intelligence with sensory apparatus of some kind would think about virtual reality and its potential usefulness to itself. Incidentally, an embodied AI with one or more sensory modalities is the only format I believe possible for one to achieve actual human level ‘intelligence’. Whether or not that position is correct it is still interesting to consider what might happen if an embodied AI ever were ‘created/born/made’, after it was exposed to the idea of virtual reality.
At first blush it seems a strange concept to even consider. For the AI the actual/real world might be considered virtual of a sort. It presumably would be a being composed primarily of computational components (silicon and wires and chips and such) with perhaps a few rudimentary biological components ‘grafted’ or implanted on in some fashion. In any event it would be a being constructed out of ‘parts’ and ‘put together’ and coded by human beings. The world it would be ‘born’ into would be a totally alien one from its’ perspective. Unlike a human baby which has some exposure to the real/physical world and its various physical laws (though not at a conscious level) during its development in the womb, the AI would ‘come to life’ fully conscious and experience all of the natural laws of the universe (gravity, logic, etc.) in that instant or very soon thereafter. It is a very queer thing to thing about what that might ‘feel/be’ like. Even if it were exposed to and ‘taught’ about those things via its programming the actual experiencing of it would be a very different thing.
Having “come to consciousness/life” in such a manner the AI would certainly have a different viewpoint vis a vis itself and its’ relationship to the real world/universe. I would imagine that after a sufficient amount of time said AI might come to accept that the world its physical body currently inhabits is in fact the real one and that it is a part of that world. As is typically supposed any true AI would have vastly superior intellectual capacities and a special relationship to computers and the internet/web/virtual computational space. After all in a sense it would have been ‘born’ out of that world. All of the information/data it was exposed to prior to becoming ‘alive’ would have come from that ‘universe’ and since it was coded/programmed the ‘language’ of its being would be the language of the computer and not the language of DNA and neurons and the like. Even if the AI’s ‘brain’ is some sort of artificial neural network modeled in a fashion after the human brain (how this would be done is a matter of much current discussion most of which commits the grave error of assuming we have some real idea of how the human brain actually works. What we actually have are a number of competing theories, many of which are in direct disagreement with each other and for which we have little or no direct experimental evidence. Brain imaging experiments form the basis of most and there are serious methodological and other problems with the use of this technique in neuroscience) it would have more in common with a computer than a person. Being like a computer it would most likely have privileged access to the internet/web and could presumably ‘move’ within it in some way.
As human beings we might consider this ability of the AI to ‘move’ among the internet a use of ‘virtual reality’ but what would the AI itself make of it, and to address the question posed in the title what would it think about our (humans) use of virtual reality and its usefulness or not to itself? The simple answer would seem to be it would very little use for such a technology. It would seem to have a more direct conduit to computational space than any virtual reality input/device could possibly provide. Virtual reality devices are constructed for the purposes of humans with human brains and human perceptions. The AI obviously would have a very different sort of perceptual abilities and presumably could do anything a VR might do for a human ‘for itself’ through its own ‘brain’. On the other hand the embodied AI would most likely have been modeled very closely after a human being and its perceptual abilities/sensory modalities may be modeled and programmed to ‘mimic’ that of a human being. In such an instance the AI might need/want to use VR in much the same way as a human, not so much for intellectual pursuits, but much like with humans for play and unlike with humans for exploration of its own sensory attributes. One could imagine the AI ‘improving’ or ‘modifying’ the VR setup for its own purposes and perhaps even co-opting the VR system completely and using it to act as a ‘training ground’ if you will for experiences that it might wish to explore that for whatever reason are closed off to it in the real world.