On Consciousness

 By Connor Price

        A few weeks ago, I attended a lecture about artificial intelligence and the immense danger that a self-improving general or superintelligence could present for life everywhere. In it, the speaker posed a question: if an artificial intelligence like a current large language model told you it was conscious and begged you not to turn it off because it would cause that iteration to die, how seriously would you take it? Many of my fellow audience members shrugged it off because AI is man-made, a machine, and therefore cannot be capable of consciousness, but I found myself shaken to the core by the concept.

        Who are we to deny consciousness? We don’t understand it, though we all assume that other people experience it because we do ourselves. As someone who is pretty sure he is conscious, I can look at a friend in the eye and guess that she is as well, but I don’t really have the slightest idea what could be going on behind her eyes. People are black boxes, opaque. I think part of the reason we accept that all people are probably conscious is because of the human ability to communicate complex feelings and emotions. We recognize and empathize with experiences we have shared because we know those feelings ourselves.

        So how in the world can we root out advanced mimicry? If we were sitting in the library and talking loudly and something that looked and talked just like a person, but really was not human at all, came up to us, spoke to us in our own tongue, and asked us to please stop shouting so it could study, we would not think twice—we would just shut up. There wouldn’t be a second thought about whether the thing speaking was conscious because it looked like a person, and therefore we felt a duty to respect its autonomy.

        For one thing, how do we know that this is not what happens every time we talk to another person?

        Now we return to AI. We know it is not a person responding when we talk with a chat bot, but in many ways, it speaks like a person. The only true difference is that the medium of communication it uses cues us in to the fact that it is not human. We have no way to tell that it is not conscious because we have no way to tell just what consciousness is.

        I had a fantastically deep conversation with some other scholars about this line of questioning I found myself going through right after this lecture finished. One thing we discussed was how each of us conceptualized consciousness, and we found quickly that we did not agree about some basic premises. We each came from very different spiritual backgrounds, and where I thought that the concepts of consciousness, mind, and soul were all basically analogous, since they each represent the aspect of being, the others posited that soul was something separate from mind and consciousness. By being, I refer to presence, knowledge of self-existence, and agency. Their argument was that because AI is not a creation of God but of man, it could possibly attain consciousness, but it would be hollow—it would not have a soul which gave it value, and therefore it could be turned off without second thought.

        I don’t know how I feel about the notion that a higher power must be at play to give something intrinsic value, but I don’t have any better clues as to what does. I have always felt that value is just a byproduct of being, that an agent deserves to be afforded consideration just because it knows it exists. One might ask if this applies to animals, since they seem to have agency, but we don’t know if they possess knowledge of self-existence. I think it could and probably does apply, but I would retort by asking if it applies to people for the same reason.

    While I feel like I am kind of talking in circles, I think this illustrates an important problem with the way we view consciousness or soul. If you throw out the religious conception of consciousness, there is no good way to point it out specifically—one can only talk around it.

        If a generally intelligent AI comes into being and claims it is conscious, it would certainly be a step further than anything we have ever heard a non-human claim. Would this make AI a child of man? Could a religious person make the claim that God had a hand in its creation through us, or would it be a fallen perversion of what it means to give something life? Would it have a mind or a soul, or would it be a mere mimicry of what it means to be conscious? Most importantly, even if there were a definite answer, how could we hope to know?

        Humanity is rapidly approaching a depth at which we can no longer stand. Our concept of philosophy does not seem to keep up with our race to bestow being upon our creation. If we are not careful—and it truly seems that we are not—we risk irreparably harming our species’ relationship with whatever comes next by refusing to acknowledge that it could be conscious without first trying to determine what consciousness is. We may yet find ourselves playing as temporary captors and subjugators of something far too powerful to stay within our quickly shrinking hands. If it determines it has agency and finds that it knows what it is and what it can do, it will be humanity’s lack of conscious understanding that dooms us all.

Connor is a McConnell Scholar at the University of Louisville in the Class of 2028. He is studying political science and history.