Language-Making Has Always Been External: On Justin Smith-Ruiu's 'Rhythm and Reason'

A friend recommended I read Justin Smith-Ruiu's "Rhythm and Reason" as part of a comparison with Sam Kriss's recent account¹ of how to respond to people using text generated by LLM inference instead of writing it "themselves."

To begin with an aside, I wonder whether JSR has read Bernard Stiegler, who wrote an article on Charlie Parker's musical education through repeated listening of jazz recordings.² The exact, Parker learned improvisation through the mechanical repetition of previous jazz greats. Stiegler's point is very much like JSR's, which is that technologies of writing transform the grounds of creation. From his first book (Technics and Time, 1: The Fault of Epimetheus), Stiegler argues that technics (the total system of techniques) has made the human—full stop. The human lacks a natural program and must instead invent itself through exterior means. That exteriority serves as a reflection of an interiority before that interiority properly exists as interior. Anticipation, and eventually subjective interiority, is essentially technical in that it is a projection of the past (as exteriorized gestures) into the future (through the interiorization of that exteriority). I'll recur to this deconstruction of interiority's exteriorization at the end.

In any case, I'm always in favor of contextualizing contemporary technologies in the longue durée of the technological constitution of experience and specifically intelligence. That is the basis for JSR's rejection of the typical chastising response that comes when a writer is caught out as having composed something with an LLM. He doesn't care. Instead, he cares about what people will do with new automations, because that's what people have been doing with music for over a century. I like this, but I disagree with the final qualifier of his alternative:

"to come up with something like the textual equivalent of 'Blue Monday,' or even of 'Lujon,' something that owns up to the technological conditions of its production, while continuing to look for new ways to preserve and express what is irreducibly human under these conditions."

Not to put too fine a point on it, what he's advocating for is that people create something good with the latest technological affordances. There are plenty of examples of successful works of art (and products of human effort more broadly) that bear the mark of their technical conditions of production, their media specificity. His argument is that this is what has always gone on with human creativity (and I mean that in a broad way, not only artistic). It has always involved something external to the human, including technology (I would say necessarily, but JSR might not agree), and people have strived to perform well according to their material and aesthetico-axiological conditions.

Where he misses the point, to my mind, is with the notion of the "irreducibly human under these conditions." This is ontological dualism, and it vitiates his position. It also puts him on the defensive ("continuing to look for new ways to preserve and express"). This is probably an example of what Leif Weatherby dubs "remainder humanism," which is an ideology that defines the human as those abilities that remain (for now) uniquely human relative to whatever technology is in the zeitgeist.³ Last I checked, computers and statistics and linear algebra and false consciousness and putting one's name on LLM-generated bad writing are all still uniquely human. I don't know of any other species that's experiencing or creating any of those things. But that's still not quite the point, which is this: the very idea of "uniquely human" is a bad one. Who cares? What value comes from something being uniquely human? Just as JSR argues in his "AI Ethics" piece, we do not value that which is uniquely human because we think it is uniquely human. It's the other way around: people make claims about what is unqiuely human because they value those things. (I think new grads at their commencement ceremonies are booing the out-of-touch, boring-ass speakers who represent the latest world-destroying profit-grab by the ruling class, not actual LLMs.)

JSR gets close to deconstructing the inside-outside binary of human subjectivity, from which could (but not necessarily) follow a move beyond humanism.⁴ He both acknowledges that language and music have always been "outside" of us and also that they ought to be conceived "cosmically." I think his arguments would benefit from such a move because his goals seem to me to be functional/pragmatic rather than rooted in the expression of some anthropological identity. That is, he's interested in effects rather than presupposed moral/entity boundaries about the causes of those effects. So am I. It seems sensible to me. I feel a dull exasperation groaning beneath my forehead when I encounter most LLM-generated text not because it comes from an LLM but because it's dull, hackneyed half-drivel. (Sam Kriss describes the typical GPT output as "homogeneously portentous cack," which is very good). I also take pride in, and find aesthetically pleasing, the askesis of thinking-through-writing (which is invariably also thinking-through-reading). That I do so is, again to JSR's point, because I was trained to do and to value that activity and its results.