The Seductiveness of AI’s Coherent Fiction

Roger Berkowitz
Babette Babich turns to Nietzsche to think about the question of why we are enthusiastic about AI. One reason, of course, is the belief that AI will "solve" our social problems. Diseases will be cured, welfare reorganized, poverty overcome. AI will solve the irresolvable social problems that we humans have not been able to. Of course, this belief in the power of AI to solve problems of human organization depends, first, on our willingness to outsource human challenges to inhuman logic, and, second, to our willingness to actually implement solutions to human problems that we humans can't understand. 

But Babich turns us to a second Nietzschean dimension of our psychological longing for AI: That it offers us the coddling and satisfying confirmation of the logicality and meaningfulness of our world. Nietzsche teaches that we prefer the lie to reality (something Arendt confirms as one driving force of totalitarianism). The lie is clean, simple and coherent, where reality is messy, complicated, and incomprehensible. What AI offers, in its non-human way, is a simplification of our already existing reality, culled from the trillions of words and sentences and phrases we have spoken and written. Nietzsche’s famous example is the leaf. We see a leaf. It may be an Oak or Maple, a Birch or a Fig. But we call all these “leaves.” Even amongst 2 or 100 Oak leaves, there are differences. And the one Maple leaf changes color in the Fall.  Yet we are happy with the lie that calls all of these leaves by the same simplifying and plurality-denying noun. As Babich glosses Nietzsche’s point, there is a necessary human comfort in the lie. Truth is a lie, but it is a lie that we need in order to live in a shared and common world. This is what she calls the “bubble effect.” 

One reason we are so enthusiastic about AI, Babich argues, is that it is so adept at confirming the lies, and the illusions that we already take comfort in. In confirming our biases and our comforting illusions, AIcan become, Babich suggests, a powerful engine of the status quo, a world that finds in the lie the coherence of reality. She writes:

 
People who have used ChatGPT enthuse about how wonderful it is — and why would they not enthuse? The effect is a bubble effect, as one may also extend what Tor Nørretranders called the ‘user illusion,’[7] the results correspond to their own data traces, littered around them in the wake of their internet use habits, reading and reinforcing the same things again and again. We like what we know; what we agree with, we like even more. We like our own formulaic expressions however primed into our consciousness as we take this to be, our own “convictions” as Nietzsche says. Thus, psycholinguists know that the key to a good conversation, to good therapy, a good job interview (or a good date) is to repeat what the other party says right back to them. Deftly done, this is not perceived as parroting but as geniality. One heart, and one soul. Perhaps most telling — this is a common feature of hype — people who have not used ChatGPT — there is a learning curve — also enthuse about it....
 
Recall Nietzsche’s Goethean reference to a leaf in On Truth and Lie, as he varies this with respect to a tree (likewise Goethean) and to the example familiar in these days of AI and screen time, regarding the cognitive physiology/psychology of reading/scanning.[31]
Throughout, Nietzsche focuses on the illusion, the deception, the lie, which, to complete the reference to AI, is automatic.  Thus I quoted Nietzsche on reading a page and here we may continue: “it is so much easier for us to put together an approximation of a tree. Even when we are involved in the most uncommon experiences we still do the same thing: we fabricate the greater part of the experience and can hardly be compelled not to contemplate some event as its ‘inventor’.” As Nietzsche emphasizes, “one is much more of an artist than one realizes.”
Thus inventive, and here we are back to the toaster we might blame for malfunctioning or to the plush toy for the elderly ladies who are given, quite as children are given, pacifiers or iPads in place of parental attention to quiet their claims on family members for such attention, or who perhaps have no such family members, ‘family’ being a phantom concept in any case, not unlike Descartes’ phantom limb.  Same diff, were this a different essay on the sex dolls that did a smash business, so I am told, during lockdown along with every other deliverable commodity item.
The point to be made — this is how one goes about constructing a usable ‘user illusion’—is that this same success led to the urgent need for robot ethics: one wants, the corporation needs, users who play by the rules. Users must never (hence, the ethical imperative) treat these products as heterosexual men have traditionally treated woman: the damages would be unimaginable in a strikingly short time. But also the user experience would be better if the user can be programmed to use the device only so and not otherwise. Both points are critical for a sustainable business model.
Beyond Goethe’s morphology — i.e., the reference to the leaf and the tree Nietzsche connects with reading and translating — Nietzsche borrowed his own text from Gustav Gerber’s Kunst der Sprache [The Art of Language]).[32] To this extent, Nietzsche/Gerber emphasize the projective element in reading a text (we think, at least to begin with and sometimes even after sustained and repeated reading, that we already know what the author is saying), which is similar to looking at a picture (especially if we have never taken a course in art history much less art, that we ‘know’ what we are looking at). Nietzsche proceeds in Beyond Good and Evil to extend our confidence that we ‘already know’ to our face-to-face interactions as Nietzsche took his insight to our perception of intimacy:
“In a lively conversation I often see before me the face of the person with whom I am speaking so clearly and subtlety determined by the thought he is expressing or which I believe has been called up in him that this degree of clarity far surpasses the power of my eyesight — so that the play of the muscles and of the expression of the eyes must have been invented by me. Probably the person was making a quite different face or none whatever” (BGE §192).
But that means, to the point of sentiment, that we make up the connection ourselves. What is more, we are so focused on ourselves that anything that seems to be focused on us has an advantage. This is the heart of Dale Carnegie’s 1936 guide for salesmen, encouraging them to learn their client’s names and to repeat them as often as possible. Repeating a person’s language is the key to ‘winning’ friends and ‘influencing’ people. AI has been doing that for a while. Nietzsche’s argument that we prefer the lie, the illusion, to truth means that when it comes to friends and lovers and family members, but also political figures, historical facts, paintings, or musical pieces, we, most people, so Nietzsche argues, “prefer the copy to the original.” We like things to be as we imagine them to be. And AI is custom-made for that.