Empires of modern passivity

A few weeks ago the New York Times ran a package of short pieces about “how TikTok has changed us,” pegged to legislation passing that could possibly ban the app in the U.S. That framing makes it seem like there is something unique about TikTok specifically, but the elements of its popularity are not mysterious: Amanda Hess sums them up as “short-form videos, algorithmic timelines and searchable sound clips,” as well as the billions of cell-phone cameras and network of millions of users that those elements depend on. The package intro also notes another important quality: “Unlike Instagram, Facebook or Snapchat, TikTok didn’t build itself around social connections. Its goal is pure, uncut entertainment.”

The other platforms have now largely followed suit, so most of them primarily incentivize the development of attention-grabbing techniques that can be isolated, copied, and applied to any kind of content with a minimum of context. Viewers can enjoy “attention-getting” as a kind of content in and of itself (perhaps that is what “pure, uncut entertainment” means) and keep current with what approaches to attention getting are effective and normalized. TikTok shows a steady stream of ostensibly ordinary people, people the audience has no particular a priori stake in, trying to be interesting, so the time spent on the platform becomes not only an ambient experience of parasociality but also an education in how to seem interesting to no one in particular. It exposes viewers to an evolving grammar of poses, postures, idioms, vocal tones, settings, editing tricks, and facial expressions that establishes a palpable sense of media presence in the absence of physical presence and garners social approval in the abstract. It thereby teaches a contemporary form of etiquette: These are the acceptable, i.e. lucrative, ways to broadcast your identity to strangers. This is how you can commodify yourself.

Even in videos that don’t feature people, a similar spectacularization is at work. One of the items in the Times package, by Becky Hughes, is about recipe videos. Where printed recipes depended on cooks reasoning their way through step-by-step instructions presented in as dry and rational a manner as possible (hence their peculiar tone; their terse imperatives), videos simply show the process and expect viewers to mimic what they see. A viral recipe is more like a viral dance craze or challenge; you participate through gestural emulation rather than through discourse.

Hughes writes:

for all the access to techniques and cuisines that TikTok has provided home cooks, the platform favors concepts over actual recipes — eggs fried in a puddle of pesto, sandwich fillings chopped into a homogenous mixture, mini pancakes served like cereal. The most shareable recipes are the ones that you can watch once, then turn around and make — no measurements, bake times or reading needed. Just dump, stir, like, follow, repeat.

I would guess that the most “sharable” recipes would be the ones you don’t need to make at all but can watch over and over again as a performance, a video poem of metamorphosis. Video recipes can be enjoyed passively, as entertaining spectacles (what Hughes here calls “concepts” ) rather than practical aids. The same suite of techniques that make people you’ll never meet seem compelling can also be applied to food, to make it seem interesting even when you can’t eat it. Whether one actually learns how to cook a particular dish is incidental to learning how to stage and consume a spectacle. You need certain ingredients and equipment and manual dexterity to make and enjoy a meal; to enjoy a spectacle, all you need is a phone.

The way TikTok conflates experience with voyeurism makes it a somewhat clear demonstration of Guy Debord’s “society of the spectacle.” Debord argues that under the conditions of late 20th century capitalism — conditions of media centricity and monopoly that have only intensified into our century — spectacle and lived experience are in a complex dialectic that sustains a generalized alienation and a universal reification. “It is not just that the relationship to commodities is now plain to see, commodities are now all that there is to see; the world we see is the world of the commodity.” Debord concludes that individuals are “condemned to the passive acceptance of an alien everyday reality” and are driven to “resorting to magical devices” to “entertain the illusion” of “reacting to this fate.” TikTok could be considered as one of those magical devices (along with the phone in its entirety) that manages that dialectic. Under the guise of “entertainment,” passivity reappears to the entertained individual as a kind of perfected agency; alienation is redeemed as the requisite precursor to consumer delectation.

TikTok makes it plain how this spectacularizing logic applies to everything: Anything at all can be profitably presented as entertainment at the expense of the specificity of its own concept. That is, anything can become entertainment instead of what it is — it is more valuable to the system (as reflected by the infrastructure provided by TikTok and other platforms) as spectacle than as a reflection of its own content. A video recipe has to entertain; it doesn’t have to help you make something edible. What’s important is that it prompts viewers to equate viewing with doing. The resulting passivity allows the cycle to strengthen itself — one is not actively seeking any content or information, so one’s impatience and expectations of being entertained continue to grow stronger.

“The spectacle is essentially tautological,” Debord writes, “for the simple reason that its means and ends are identical. It is the sun that never sets on the empire of modern passivity. It covers the entire globe, basking in the perpetual warmth of its own glory.”

Open AI’s Chat-GPT4o — a new iteration of its chatbot product that emphasizes its ability to mimic personability — can also be understood in these terms. Its means are its ends. It subordinates content to the interface, information to the way it is presented, and works to marginalize its significance so that the purpose of chatting with a bot is to keep chatting with a bot. Max Read concludes from Open AI’s recent demo that it has realized that “the point of A.I. is to talk to a cool computer.” It’s an interface that customers can use not to get anything done but solely in order to enjoy an interface, in a techy approximation of why people chitchat.

It’s hard not to see OpenAI’s anthropomorphic turn as an admission that mimicry is all that chatbots are good for — part of a pivot away from productivity and labor automation and toward selling pseudo-companionship, automating away your need for friends. (This typically credulous Kevin Roose piece, in which he recounts his experience testing out some AI companion products, seems indicative of the industry’s drift into the experience economy.) Read quotes John Herrman, who points out that “what OpenAI presented was instead primarily a step forward in its products’ ability to perform the part of an intelligent machine,” and not some step toward competence or factual reliability. Facts are irrelevant to the truth of the experience OpenAI sells, which is grounded in the fantasy of having someone you can always talk to and boss around without having to worry about managing their performance. The chatbot performs “assistance” in a way that makes its actually doing anything more and more beside the point, much as the recipe videos make cooking anything beside the point. Users can just consume the entertainment value of a chatbot’s responsiveness as a reified phenomenon — what Paul Ford describes as generative AI’s “shamelessness”:

What I love, more than anything, is the quality that makes AI such a disaster: If it sees a space, it will fill it—with nonsense, with imagined fact, with links to fake websites. It possesses an absolute willingness to spout foolishness, balanced only by its carefree attitude toward plagiarism.

Models are trained to sound confident, and that performed confidence can be enjoyed separately from what that confidence is supposed to be attached to.

For his part, Read also likens generative AI to a confidence game: magic tricks. He writes:

a few years in to the generative-A.I. craze, it seems pretty clear to me that these apps, in their current instantiation, are best thought of, like magic tricks, as a form of entertainment. They produce entertainments, yes — images, audio, video, text, shitposts — but they also are entertainments themselves. Interactions with chatbots like GPT-4o may be incidentally informative or productive, but they are chiefly meant to be entertaining, hence the focus on spookily impressive but useless frippery like emotional affect.

Where TikTok instilled a passivity that is limited to spectatorship, chatbots render conversation (normally an active, reciprocal process of engagement and potential conflict with another consciousness) into something passive and consumable while still taking on the flattering trappings of participatory activity. Just as the point of TikTok’s algorithmic feed is to get people to consume more TikTok, so the point of a chatbot’s conversational interface will be to get people to keep talking to it. Who wouldn’t want to have a chat with a talking advertisement? It is like an algorithmic feed that can put into words how everything is centered on you. Even if it could reliably do anything else, it wouldn’t feel as satisfying as that.