maybe I can start with some stray thoughts on autonomy
What to make of autonomy? This has been on my mind again recently, after a year’s hiatus post-MFA. One hears much online and in all sorts of essays on agency and autonomy. Although not explicitly on this topic, I recently read a blog posting by Alexander R. Galloway in which he posits the notion of whether whether algorithms are biased (Are Algorithms Biased?). While taking pains to preserve their distinction from each other, he makes connections among algorithms, math, logic, programming languages, software, tech, and AI, and briefly discusses some of their failures along the lines of race and gender. More directly, however, Galloway goes over what seems to be so much resistance against any such endeavors at “the politicization of math and code.”
This is certainly a topic of interest and has been for some time: I recall, for example, an old conversation some 25 or 30 years ago between me and a good friend, an analytic philosopher and specialist in Ancient Greek and Roman philosophy, who balked at the notion of an inherent bias in logic. She also happens to be a woman, which may or may not come as a surprise. Nevertheless, I am more interested in an adjacent issue, one that the aforementioned discussion seems, in my mind, to hinge upon; and that is whether algorithms may be (or become) autonomous. Indeed, when/if they do possess some level of autonomy, to what degree might it extend? And if they do not now have any sort of autonomy, would it nevertheless aid in any sort of discussion to imagine that they do possess it?
Here, I am perhaps conflating autonomy from bias. If an object of creation – whether painting or equation, building or algorithm – is biased, it stems from its direct connection to its creator(s) and/or the conditions of its creation. What characterized the creator inevitably colors the creation. It is a mere characteristic of that object. To say that an object is autonomous is to say that it is free of that sort of enduring, perhaps inextricable, connection, that it is in fact unbiased and independent; but it is also might be saying something more. It seems to suggest that it may have some sort of volition all its own. Is there more to the connection between bias and autonomy?
I link bias to autonomy intuitively and explicitly in part by recalling Stuart Hall and his notions of the encoding of biases into media (i.e. Hall, “Encoding/Decoding,” Ch. 10 in Culture, Media, Lanuguage). Hall’s notion of racism explored throughout his writings is not simply an issue of various individuals’ biases toward various groups, but a basic element of the economic and social structure, the very way we experience and understand our cultures and ourselves, and thus an integral part of economic (and military) power: it is not only in the overt ways our enemies are chosen, it is also in the gasoline used for the cars we drive; it is in the cocoa and the sugar of the chocolate we eat; it is in the rare-earth elements dangerously mined and the labor exploited for the digital components we keep in our pockets. While pervasive and structural, such biases are also typically not obvious or transparent. In this way, such bias comes unlinked from and individual’s conscious or unconscious racial animus or intentionality; and any and all objects of intentionality may thus be interpreted to express things never dreamt by the individual.
On the one hand, there is the writer or artist who creates a work; and on the other, there is the reader, the viewer, the user who perceives, experiences, and then interprets that work if they so choose. Between them, there is the work. Although perhaps an object, whether material or intellectual, a work of art is not like any other object in the world, for it is an object of intention, rendered by the mind and/or hands (body) of its creator. It is thus fundamentally different from a rock in a stream, a grain of sand on a beach, a fern in a jungle, or an iceberg in an ocean. Those objects are ever available to the perceptual experiences of their beholders, but there is no hermeneutic of the natural world. That fern may be green or that rock may be smooth, and while perhaps up for geological or botanical analysis, they cannot be interpreted. They do not mean anything in and of themselves, for they simply exist.
Or do they? What about culture and language? Is a jungle or an iceberg ever simply a jungle or an iceberg; do they really have no meaning, especially with regard to today’s simultaneous disregard and preoccupation with global warming? A rock was not even a rock until was designated as such, for that word does not capture everything that it is, that it has been, or that it will ever be. The object has been named, and is thus delimited. If, as I have here previously alluded, an individual’s autonomy is limited by their inability to completely control the explicitly or implicitly biased meanings of their objects or intentional acts, what is there to say about those intentional objects once they have been released? There is the creator of the work, the receiver of the work, and there is the work in between; all of them, however, exist within culture and language. What sort of limitations have thus been placed upon them by that culture? Are their cultural constraints collaged, bonded, handcuffed, tied, tethered, sewn, woven, or welded upon them as immovable weights? Or are they not constraints at all, but wings?
If an algorithm is unbiased, that seems to suggest that it has been liberated and is thus autonomous of its human origins. Alternatively, it nevertheless also suggests the transcendence of a less visible enclosure composed of a culture’s economic and social systems, so pervasive as to be all but invisible, and working its influence through everything. Is that a possibility? Is it even desirable? Are humans themselves even autonomous?
Is there any sort of interest or advantage to imagine a partial, or an incomplete autonomy?