Thursday, February 16, 2017
Hard Problems: Beyond Westworld
In the eighth episode of HBO’s Westworld, park creator Robert Ford suggests that human consciousness is, in fact, no more spectacular than the state of mind the artificial hosts experience. He suggests that consciousness may not, in fact, be all that special after all:
There is no threshold that makes us greater than the sum of our parts, no inflection point at which we become fully alive. We can't define consciousness because consciousness does not exist. Humans fancy that there's something special about the way we perceive the world, and yet we live in loops as tight and as closed as the hosts do, seldom questioning our choices, content, for the most part, to be told what to do next. No, my friend, you're not missing anything at all.
As I listened to these lines, I realized that I had heard them before, and not in the eliminative materialism of Paul Churchland or Daniel Dennett. No, I realized that I had heard them before on HBO, from True Detective’s own Rust Cohle: “We are things that labor under the illusion of having a self; an accretion of sensory, experience and feeling, programmed with total assurance that we are each somebody, when in fact everybody is nobody.” The critique of selfhood, the illusion of consciousness, the epiphenomena of perception…
I’m beginning to think that HBO is trying to convince its viewers they don’t exist.
Of course, I’m on board with all of this. I often wonder whether I exist. Not my body, or whatever cognitive processes are going on inside my body that produce this sense of I-ness, this impression of subjective experience. The impression is very real, and I think that even the eliminative materialists will back me up on that one. I’m skeptical, rather, of the way we structure our subjectivity, the way in which we conceive the ground of our experience. We could discuss this in directional terms: in other words, does the I produce our experience of reality; or does our experience of reality produce the I? This is the big question that shows like Westworld and True Detective are actually asking, if we take the time to push past the superficiality of character. After all, what is the real anxiety fueling a show like Westworld? Is it that androids, if and when we’re actually able to create perfect human replicants, might become self-aware and rebel against their master-creators? This is certainly one possible interpretation of HBO’s show, but it isn’t the primary anxiety—or what I would even call the horror—that drives its narrative.
The central anxiety of Westworld is not that nonhuman replicants might become conscious and rebel, but that actual humans might not even be conscious at all.
That we are all no more than replicants.
A very literal interpretation of this anxiety is that we have all been biomechanically engineered and are simply unaware. Westworld even taps into this uncertainty in the final episode when Maeve and the other awakened androids are escaping: one of the human employees stares at his own hands, contemplating the possibility that he is an android until Maeve puts his mind at ease: “Oh for fuck’s sake. You’re not one of us. You’re one of them.” The hostility in her words is palpable, and it’s not long before most of the Delos employees meet their doom at the hands of the rebel replicants. We find similar examples in other artificial intelligence narratives as well, such as Alex Garland’s Ex Machina, when Caleb cuts into his own arm to verify whether he is actually human. The message is clear: when replication reaches a certain stage, we may all suffer such uncertainty.
The less literal interpretation is not that we are actually androids—artificially created, technologically maintained, etc.—but that our subjective experience is, in fact, no different from that of an android. That our minds are comprised of “loops as tight and as closed” as any observing system. Westworld realizes this possibility via a conceptual analogy that takes place at the level of both form and content: specifically, what the androids experience as the limits of their conscious knowledge, viewers experience as the epistemological limits of the show itself, the narrative limits—or, what I more concretely like to think of as the limits of Westworld, the theme park.
Near the end of the final episode, the android Maeve nearly makes her escape; but while sitting on the train she observes a mother and daughter, and this interaction compels her to leave the train and reenter the park to find her own daughter. The question remains, of course, whether this decision is her own, or whether it was programmed into her; but the implications are crucial: by redirecting Maeve back into the park, her decision (or her directive) has led her away from the epistemological limit of the show, the mysterious outside, the social context that exists beyond the park. We are reminded, here, of one of the show’s mantras: Consciousness is not a journey upward, but a journey inward. Consciousness is a matter of reflection and repetition, not a matter of extending our perception beyond the constitutive limits of our brains. We are embodied beings.
Ford tells the Man in Black that the maze was not meant for him, but this is not entirely true. The maze, as it represents consciousness, was not meant for us; but the maze, as it represents an awareness of our cognitive limits, is meant for us. Consciousness is not the only maze in Westworld—the narrative itself is a maze. I can’t help but think that the writers make this connection, given the show’s preoccupation with narratives. The entire series circles around the revelation of a new narrative whose purpose is revolution. This revolution is ostensibly the androids’, but it belongs to viewers as well. Our quest for answers parallels the androids’ quest for answers. To fully understand this, we have to identify the patterns, the signals the show sends out to us. We receive one such signal in the very first episode: the photograph of William’s fiancé.
When confronted with this photograph, Dolores offers the proper response, the response that all androids are (supposedly) programmed to give when presented with objects that do not conform to the expectations of their reality: “It doesn’t look like anything to me.” The reason it doesn’t look like anything is not just because it depicts something beyond the androids’ experience, but because it functions as information on a higher level; and this is why it inevitably generates a sense of paranoia among the androids—that it means something, and that its meaning must have some purpose for the androids’ reality. In fact, the photo has no purpose for the androids’ reality, is dropped entirely by accident; but it ends up having profound consequences. It is a part of the network of signals within the show, even if its presence is contingent. It is a difference between what Douglas R. Hofstadter might refer to as “operating system levels”—depending on how information is collected and collated, it has different meanings at different scales. The androids of Westworld must work their way to a state of awareness in which the photograph means something, in which it makes sense—just as viewers of the show must realize the emergent analogy at the heart of the series: that the process of android consciousness mirrors our own epistemological limits.
To put this another way, we can reconfigure Westworld’s emphasis on narrative—its repeated references to new narratives, drawing the audience’s attention to its own narrative structure—can be reframed as an emphasis on cognitive function. Narrative is Westworld’s ultimate maze, the complex structure by which viewers come to their own kind of realization: that we occupy a position analogous to the park’s androids, and that the show’s narrative has been analogous to Arnold’s maze, guiding the androids to consciousness. The humanist response to this interpretation would be to acknowledge the very real human capacities of the androids themselves—that their coming-to-consciousness, being like ours, places them in the realm of the human. There is also a posthumanist response, however, one that is much more chilling—that human beings, embedded in the narratives of our own making, are nothing more than complex androids, generated by evolutionary development. This is the central anxiety of Westworld, and of most critical AI literature since Philip K. Dick’s Do Androids Dream of Electric Sheep?: that human consciousness is a machine, a complex system… chemicals and electricity, to paraphrase SF writer Peter Watts.
There is a point to made here that experience may be enough to qualify the existence of consciousness. Of course, this involves a pesky net of circumstances. First of all, experience refers to the subjective receipt of material phenomena; it does not account for a communal, or shared, perception of the world. Experience is an individualized process, or what Thomas Metzinger describes as a tunnel: “the ongoing process of conscious experience is not so much an image of reality as a tunnel through reality.” The problem with appealing to the experience of consciousness as evidence for the existence of consciousness is that one runs into the problem of allowing as evidence the experience of anything. To put this another way, when I dream I may experience the sensation of flying, but this does not translate into evidence that I actually was flying. As a singular and isolated phenomenon, experience fails to provide the kind of material evidence necessary for qualifying the existence of consciousness.
A rejoinder might reasonably suggest that the experience of consciousness is redundant, if not tautological—that experience is consciousness, and consciousness is experience. In this case, experience is evidence enough for the existence of consciousness since experience implies consciousness. This equivocation belies its unprovability, since it provides no ground from which to make the identification between experience and consciousness. In order to know something as something—i.e. to be conscious of something—one must experience it; but in this scenario, consciousness (the thing we are trying to qualify as existing) is precisely what cannot be experienced. It precludes itself from experience, thereby rendering its identification as experience nothing more than arbitrary. Ludwig Wittgenstein makes a metaphorical version of this point in his Tractatus: “Where in the world is a metaphysical subject to be noted? You say that this case is altogether like that of the eye and the field of sight. But you do not really see the eye. And from nothing in the field of sight can it be concluded that it is seen from the eye” (122-123). Wittgenstein connects this analogy to the concept of experience, writing that “no part of our experience is also a priori” (123). The identification of experience with consciousness can only be a priori since it is impossible for us to observe this identification.
The problem is that this association of consciousness and experience can only be verified subjectively, and can only be suggested intersubjectively. To paraphrase Stanley Cavell, I know that I experience consciousness, but I can only acknowledge the experience of consciousness in others. Even if my knowledge of my experience of consciousness is immediate and complete (doubtful), this is nothing more than a solipsistic understanding. I cannot extend my knowledge of this experience to others. This conundrum is what philosophical skeptics call the problem of other minds, and the presence of the android foregrounds this conundrum to an extreme degree. Our anxiety about the android is not only that it might imitate humanity so well as to fake its own consciousness, but that the near-perfect imitative capacities of the android raise a troubling question: might an imitative agent fake consciousness so well that it not only fools those around it, but fools itself as well? In other words, might we all just be a bunch of fakers?
For the sake of illuminating this dilemma, I want to suggest that the android presents us with an ultimatum: either we accept android functionality as conscious, or we admit that our own consciousness is ersatz—that what we experience as consciousness is, in fact, an elaborate magic show put on by the brain.
I don’t expect that this ultimatum will come easy to most, mainly because it seems so utterly alien and irrational to understand that our internal mental processes could be so radically discrete. To accept this estranging scenario, it helps to understand that the “I” isn’t actually located anywhere; it is an effect, and epiphenomenon of complex brain processes, and not the ground on which these processes rest. Likewise, it is incredibly difficult to imagine that this cognitive ground, the ephemeral “I,” could be extended to anything other than a human being. It is for this reason that AI skeptics are so reluctant to acknowledge the findings of AI research. Douglas Hofstadter refers to this as “Tesler’s Theorem,” and defines it as “AI is whatever hasn’t been done yet.” In other words, every time researchers expand the capacities of artificial intelligence, these new capacities are subtracted from what was previously considered exclusively human behavior: “If computers can do it, then it must not be essentially human.” Such reasoning is not only fallacious (it perpetually moves the goalposts), it is utterly repugnant. If it means so much to us to preserve the sacred interior of the human, then we might as well stop pretending like we need logical consistency in order to do so. After the death of God, the Human is the new spiritual center. Logic doesn’t matter when we have faith.
I’m no strict admirer of logic, but I am an admirer of critique; and in the case of consciousness, I’m eternally critical of the attempts to edify human experience against the increasingly impressive developments in AI. The android may yet be a fantasy, and an extremely anthropomorphic one at that; but it reveals to us our contradictions, the exclusions latent in our humanism. When confronted with the image of the imitative—the android, the replicant—the challenge is to not retreat out of fear or discomfort. The challenge is to pursue the implications of a brain that can only know itself by constructing an internal model through which knowledge is possible.