Aesthetic Taste in the AI era: emotional labor, executive authority, and bourgeois good taste without the bourgeoisie
As The New Yorker’s Kyle Chayka has discussed in one of his recent columns, the AI bros are really into the idea of aesthetic taste. In the extreme information overload that generative AI provides, taste is the capacity to discern, and thus, to lead. Sari Azout, writing on the company blog for her upmarket Pinterest platform Sublime, argues that
Slop is everywhere, but those with taste can separate the wheat from the chaff and leverage AI to rise above the rest. Similarly, software engineer and strategist Daniel Bentes writes that “AI…cannot create a new taste…Humans, with their flawed, subjective, and deeply emotional nature, are uniquely positioned to lead these shifts.” At bottom AI is a probability generator, and it can only locate the most probable outcome its training data suggests; it cannot point to improbable or “disruptive” options. For Bentes, the human sense of aesthetic taste, with its ability to identify unexpected but pleasurable outcomes, transcends merely machinic probability.
According to former Dean of Harvard Business School Nitin Nohira, taste is also an ability that puts you up and above other people: “For leaders or aspiring leaders of any type, taste is a competitive advantage, even an existential necessity.” Defining taste as “the ability to elevate utility with elegance,” Nohira thinks this je-ne-sais-quoi is what elevates truly great corporate leadership above the merely basic; it’s what “made Apple more than a tech company.” As this example of Apple suggests, Nohira is using “taste” as a proxy for class: long the more expensive and stylish option over its PC competition, Apple is a status symbol, a proxy for upper-class status.
This use of “taste” in the age of AI remakes bourgeois good taste without the bourgeoisie and its pesky commitment to civil society/the public sphere/formal equity/etc. As I have written about before, traditional notions of taste in Western philosophy tie it to the capacity for “subjective universality,” or the ability to individually and subjectively (with feeling or judgment, not reason) arrive at the same, universal conclusions as everyone (or “everyone” in heavy scare quotes, because the whole point is that not everyone counts as someone) else. That is how “we” can all be independent individuals making our own decisions yet all coexist under universals. In this framework, to exhibit taste is to feel that fine art is superior to entertainment, for example. Demonstrations of tasteful judgment proved one’s qualification for participation in civil society, as they were evidence that you belonged to the “we” who was the “everybody” who subjectively arrives at the same universal conclusions (white women, non-white people, the working classes, with all their sentimental, excessive, and lowbrow preferences, did not qualify). Bourgeois good taste was absolutely a mechanism of gender, race, and class stratification, just one that routed through classically liberal notions of the public/private split.
Today’s discussions of taste and AI take the “we” out of the equation and focus exclusively on the preferences and judgment of private individuals. For Azout, “taste can transcend universal preferences to embrace imperfection,” such as Japanese katsougi or Pollock-ian abstract expressionism. Or, as Nohira puts it,
Taste is born of human discretion—of growing up in particular places, being exposed to particular cultural references, developing a point of view that is inseparable from personality. In other words, taste is the human fingerprint on decision making. It is deeply personal and profoundly social.
Transcending universals and the result of individual idiosyncrasies, taste is as distinctive and personal as a fingerprint. Taste is purely subjective preference, a private interiority developed from living the unique life each of us lives.
Pure subjectivity sans universality, this new sense of aesthetic taste nevertheless shares a number of features with traditional Enlightenment taste. First, it’s an imperative. As Nohira puts it, “taste is the instinct that tells us not just what can be done, but what should be done.” Kant’s idea of beauty is an aesthetic corollary to his theory of the categorical imperative. The first formulation of the categorical imperative holds that one should act in a way such that anyone else in that same situation would act – your action as an individual should be universalizable. Similarly, Kant argues that when one pronounces something to be beautiful, he (it’s explicitly a he, the gendered language here is intentional) claims that everyone else ought to think so too – his judgment is universalizable. In this 21st century context, taste retains its imperative but lacks its universality. For Nohira, taste is “the human fingerprint on decision making” that shapes things like “brand identity” and company “culture” – it is, in other words, the executive choice of a private individual. The imperative comes from his authority as an executive, not an appeal to some, erm, categorical imperative.
Like Kant’s theory of genius, 21st century aesthetic taste exhibits a kind of newness that feels like it’s natural and spontaneous even though it may in practice be the result of hard work. For example, Nohira claims that “Taste is neither algorithmic nor accidental…It has the power to remix elements and bring about plausible and even creative new combinations” – taste is somewhere between rules and randomness. Azout offers a similar thought, that taste is about “know[ing] when to break rules in ways no theory explains.” (Her use of jazz improv as a metaphor for this rule-breaking could use some reading through Dale Chapman’s critique of the way such metaphors individualize and neoliberalize Black radical aesthetics.) Bentes describes taste as “the spark of intuition and serendipity” that is unique to “human creativity.” Taste departs from existing rules, but in a way that feels fitting and pleasurable rather than anarchic and unsettling.
This non-mechanical but nevertheless comprehensible newness and originality is also a feature Kant attributes to fine art (i.e., the thing that taste would judge as beautiful rather than merely charming). As Kant explains, judgments of taste are categorically distinct from scientific reasoning:
There is no science of the beautiful, but only critique; and there is no fine science, but only fine art. For in a science of the beautiful, whether or not something should be considered beautiful would have to be judged scientifically, i.e., through bases of proof, so that if a judgment about beauty belonged to science then it would not be a judgment of taste” (Third Critique, 172).
Claude and Suno were a long ways away back in the 1790s, but Kant like the AI bros thinks taste is the exercise of a capacity that goes beyond mere scientific logic. For Kant, scientific proofs are algorithms of a sort, rule-bound processes that don’t really allow for improvisation or creativity (or what Kant would call the “free play of the imagination and the understanding” – science is just about the understanding). Some art, however, “must seem as free from all constraint of chosen rules AS IF it were a product of mere nature” (173). Taste is the ability to discern “free from all constraint of chosen rules” as if you are relying just on some inner je ne sais quois, or your personal nature. Taste looks and feels like the spontaneous exercise of unbounded freedom, even if it has to be the result of lots of practice and cultivation. And for Kant this “unbounded freedom” is crucial: the freedom of fine art and judgments of taste is one and the same with the freedom of the liberal/bourgeois subject in civil society.
AI bros use taste to similarly assert the difference between “freedom” and unfreedom, they just conceive of freedom differently than Kant did. They upgrade the civil freedom of the liberal subject into the frictionless mastery of the executive decision-maker presiding over bots and grunts that do the actual high-friction labor of working and thinking. Nohira highlights how the capacity for taste is “an existential necessity…for leaders or aspiring leaders of any type” because it is “in everyone’s interest, even people who are not at the top of the decision tree, for leaders to be able to make the right choices in the AI era.“ The capacity for judgments of taste is not just what sets humans apart from and above the mere bot, it‘s also what sets some men above others and makes them “leaders”. As Nohira puts it, “When every option is instantly available, when every variation is possible, the person who knows which one to choose becomes even more valuable.” Here, the AI bro analysis of taste leaps over Kant and looks more to Plato, who thinks that freedom is following the best master (in his case, the metaphysical True/Good/Beautiful). To be free, an individual must align himself with and subordinate himself to that metaphysical Truth, and such capacity for self-mastery is the thing that qualifies one to be master over others who supposedly lack that capacity in themselves. Nohira gestures towards a similar structure, but situates it in the context of racial capitalism (he’s talking about CEOs after all) rather than metaphysical idealism. To demonstrate taste is to demonstrate one’s alignment with the patriarchal racial capitalist order (because that is what AI reproduces, in the end – it re-makes speculative realities in the shape of our white supremacist capitalist patriarchal one), and that alignment is what is necessary to rule over others because it’s the thing that allows you to “discern” between, say, AI hallucinations that are aligned with this politics and those that aren’t.
Kant gets closest to this in his discussion of genius, or the ability to produce tasteful works of fine art, as that’s one place where he gestures back towards classical Western thought. The product of freedom,
Genius is a talent for producing something for which no determinate rule can be given…the products of genius must also be models, i.e., they must be exemplary; hence, though they do not themselves arise through imitation, still they must serve others for this, i.e., as a standard or rule by which to judge.
Just as Plato’s free man aligns himself with the metaphysically true, the work of genius is a model for others to use as their guide in their own attempts at reproducing that act. This connection to reproduction is why Kant brings up the classical Latin sense of the term “genius”: “the word genius is derived from [Latin] GENIUS, which means the guardian and guiding spirit that each person is given as his own at birth” (175). As Christine Battersby demonstrated in her 1988 book Gender and Genius, this “guardian or guiding spirit” was explicitly identified with sperm and patrilinearity. The exemplary status of the genius is the same as the exemplary status of the patriarch: he is both model for and authority over his progeny.
AI bros use their idea of taste to grant executive decision-makers the authority of the patriarch by masculinizing stereotypically feminine tasks and traits like emotional labor and aesthetic receptivity. For example, Bentes argues that “AI can analyze metadata, historical context, and user preferences, but it cannot genuinely feel the weight of history or the resonance of cultural significance” (emphasis mine). Taste accesses something below and beyond even vibes, an authentic receptivity to sensory, perhaps embodied perception.
Throughout the modern era, something like this capacity for “genuine feeling” has served as a mark of white masculine superiority because it evinces one’s ability to both (1) master bodily, emotional experiences that white women and non-whites were supposedly unable to control in themselves, an (2) transcend the limits of stereotypical white masculinity and its purportedly rational, intellectual, disembodied nature. As Kyla Schuller argues in The Biopolitics of Feeling, in the 19th century “impressability,” or “the capacity of a substance to receive impressions from external objects that thereby change its characteristics” (7) was thought to be evidence of “civilization” and refined evolutionary development: Impressibility came to prominence as a key measure for racially and sexually differentiating the refined, sensitive, and civilized subject who was embedded in time and capable of progress, and in need of protection, from the coarse, rigid, and savage elements of the population suspended in the eternal state of flesh and lingering on as unwanted remnants of prehistory” (8). White bourgeois people were thought to be able to perceive sensory information more fully and channel that perception into productive outcomes, whereas non-white people were thought to be merely subject to crude sensations.
Similarly, European philosophical aesthetics thought that elite white men could exhibit the capacity for “receptivity” that white masculinity’s stereotypical rationality and universality alienates most white men from, and which white women and non-white people can access but cannot master or control. As Robert Gooding-Williams puts it in his reading of Nietzsche’s Thus Spoke Zarathustra,
Gendering the putatively suprasensible subject of knowledge as the hero Thesus, Zarathustra announces that this subject can regain his ability to sense, feel, and revalue his passions only if he repudiates his sublime, heroic will to truth and knowledge and acknowledges his Ariadnean power to be sensibly affected by passion and desire” (48).
This capacity to appropriate feminized (and non-white – as I argue here it’s important Ariadne is from Naxos) receptivity is one thing that distinguishes the Nietzschean overman from the rest of us. As Gooding-Williams shows later in that quoted piece, in the 20th century this white masculine capacity for receptivity shifts from an ideal of bourgeois good taste to pop cultural cool as the object of appropriation shifts explicitly to Blackness. White men can’t jump, but some have moves like Jagger, etc etc. For nearly 300 years, this capacity for refined aesthetic discernment has been what distinguished elite white (bourgeois) men from both their merely rational counterparts in civil society and the irrational, subjective white women and non-white people relegated to the private sphere.
AI bros turn to taste to similarly masculinize the kinds of emotional, feel-y labor that automation has gendered feminine in new, tech-specific ways. Musicologist Catherine Provenzano has shown how automation exacerbates gendered differences in the perceived value of musical labor. Pitch correction software like AutoTune automates the technical dimensions of vocal performance by taking the job of staying in tune from the singer and putting it “in the box” (industry jargon for digital audio workstations). As Provenzano argues,
Automated pitch-correction shifts the kind of labor singers to from vocal to emotional performance. Through interviews with producers and analyses of fan discourse, Provenzano finds that in this new automated context, women’ s emotional performances are thought to be unrefined things in need of domestication by (typically men) producers, whereas men’ s emotional performances are praised as innovative:
female voices have been most subject to “taming,”…meanwhile, male voices in pop are tools for fashioning new paradigms of sung emotionality, and often an emotionality that is audibly mediated through pitch correction software” (80).
Whereas Britney Spears is shamed for using Auto-Tune, T-Pain and Future are treated as innovators who use new technology to push musical expression to new, post- or trans-human levels. In this context, automation is a tool that distills feminine things to their fundamental unruliness while also facilitating masculine things’ ability to discern and perform even more refined emotionality.
AI bros’ use the idea of taste to do the same thing: automation relieves us of the technical authorial labor of things like writing an email, chatting up women on a dating site, or texting with our parents, and thus frees men up to focus on the ineffable, highly-refined tone, tenor, or feeling of an output. When people use their aesthetic judgment to criticize the poor quality of AI slop, they are described as “bitching” and “whining” — two traditionally feminized figures of unruly emotion. But when they use that discernment to make decisions that are aligned with Silicon Valley’s weird hypermasculinity and/or business interests, that’s evidence of their capacity to lead others. In the 2020s, the idea of aesthetic taste is doing what it has always done: legitimate the authority of (mostly white) men from the economic ruling class. This time, that authority is not that of access to civil personhood, enfranchisement, at so on, but of the private-sphere authority of the CEO and the patriarch.