Slop

The term ‘slop’, which many use to characterize AI, is worth considering. ‘Slop’ is uniform, formless, tasteless food, produced in great quantity, with low nutritional value, designed for swift mass delivery to and consumption by large groups for whose specific members one holds no meaningful regard and with whom one desires no sustained interactions. ‘Slop’ is typically served to animals, to prisoners, and to other institutionalized persons, to groups that are entirely dependent, to those who have no other choice or alternative, and to those who have appetite but no taste. ‘Slop’ has much less value when there are realistic alternatives on the menu, or a menu in the first place.

‘Slop’ has caught on as a term to describe much of the AI-generated content that we encounter online, and increasingly also off. The term feels like an apt one to describe the deluge of bland, generic, uncanny, mindless, impersonal, and insubstantial content that is filling our feeds. The ease of production with AI has made it possible for lazy, untalented, thoughtless, uncaring, and dishonest people to produce innumerable shallow semblances of things that might once have reliably indicated skill, thought, and care. As our social and communicative spaces are flooded with such AI-generations, the humanity that we value in art and writing will steadily get substituted for by hollow simulations of it. Calling such AI generations ‘slop’ is a way of naming the adulteration that they involve, like a stealthy replacement of good food with a cheap substitute that does not taste entirely dissimilar, yet which lacks all the nutrients.

Such content can sometimes appear indistinguishable from material thoughtfully and lovingly created by human beings, and those who produce it will often try to pass it off as such. The recognition that it is in fact slop often feels like an insult, like knowing that the friend that has invited you to a meal is trying to pass off a cheap microwave dish as something they labored over themselves. While the quality of the food may be inferior, the real issue is the lazy avoidance, coupled with the deceptive pretense, of the generous work of preparing a meal for you. Had they been forthright that they were serving microwaved food, you might have been a bit disappointed that they did not go to more effort, but through the deception they desire the appreciation you would have given for a home-cooked dish. ‘Yet it might all taste the same!’ This is badly to miss the point.

When people receive a personal message, enjoy a creation, or have a conversation, they feel some sense of being valued. Flooding our social communications and our realms of creation with empty semblances of these acts will have a Gresham’s Law-like effect: debased communications will drive out good. A sense of mutually-accorded dignity and trust is replaced with a sense of being used and growing cynicism and suspicion. Had we no expectation of truly human communication to be betrayed, so much AI use would be of little value to us. It is only in the deceitful ‘value’ it enjoys as it masquerades as human that it will be appreciated.

Yet, as it becomes pervasive, all communication will be diminished in value and distrusted. To use a food metaphor not unrelated to ‘slop’, it is akin to ultra-processed foods, which may be engineered to be hyperpalatable, to fool us that we are eating something different, and/or to replace healthy and nutritious ingredients with cheaper substitutes. Such a product may successfully deceive many consumers that they are eating some quality and unprocessed food; yet this does not make it interchangeable with or substitutable for that food! Ingredients and the healthiness of processes of preparation matter immensely, which is why there are regulatory bodies to protect the public from the adulteration of their food.

There are innumerable applications of AI, many of them worthwhile (chiefly in those areas where they are most clearly distinguished from human beings, such as in predicting the structures of proteins or registering significant patterns in vast quantities of astronomical data). However, so many of its most prominent and popular uses, as they concern activities and creations that are more properly human, systemically encourage deception (of others and of the user themselves), corner-cutting, laziness, and disregard for others. When uses of AI substitute for or simulate communication with others, they encourage both neglect and pollution of our social commons, allowing the non-human to become ubiquitous in those areas where we most need to practice our shared humanity. As Proverbs 18:21 teaches, ‘Death and life are in the power of the tongue, and those who love it will eat its fruits.’ Protecting the relationship between our words and human interiority from being simulated or adulterated should be a matter of grave social concern.

This is a systemic problem, which cannot merely be dealt with by virtuous personal use: like the use of performance-enhancing drugs in sports, it requires strict regulation, along with, in certain contexts, stigma and punishment. As in a sport afflicted by drug cheats, the integrity of entire realms of communication and social practice—the university being a good example—depend upon ensuring that what we are engaged in is authentically and unadulteratedly human.

When establishing and applying drug guidelines in sports, there will be complicated cases, substances, and practices, which will require prudence and considered yet contestable judgment calls. Some practices may not technically cross the line of illegality, yet should be discouraged as contrary to the spirit of the activity. Different organizations will likely have slightly different rules.

However, many of the issues are society-wide, and the task of managing them cannot merely be given to private and voluntary organizations. People can often focus on edge cases and not consider the broader effects of weak standards: lowering the bar for some supposedly worthy instance also admits a great many unworthy ones. They consider how they might function virtuously as individuals, but give little thought to population-wide tendencies within radically transformed incentive structures and environments. Or they regard new technologies as things that users can straightforwardly stand over against, without considering what facilities might be lost in persons formed by them. As Antón Barba-Kay recently observed, commenting upon the papal encyclical, Magnifica Humanitas:

The encyclical does draw a categorical distinction between AI and human intelligence. It assumes and defends our capacities for “inner freedom,” responsibility, and critical thought. Yet these capacities, like our capacity for reason itself, are not fully given in advance. They are developed through specific media and ordinary practices, some of which tend to establish attitudes of attention, reverence, care, and insight. And if it would be wrong to think that a human soul could be completely frittered away by distraction, we have also seen enough to realize that it can be deformed by practices that abridge or automate thought.

Many think of the technology as if merely additive, without considering its long-term ecological effects. For instance, few consider the way that the growing shadow of government upon our daily lives—I here recall A.J.P. Taylor’s famous description of pre-1914 England—is in large measure a result of modern technologies (often of war), techniques, and systems that have subjected formerly largely self-regulating societies to disruptive forces beyond their effective control. Modern policing is down the road from the automobile. Immigration regimes need to develop to deal with the large-scale movements and dislocations of people in a hyper-connected world, wrestling with the ways that mass movements of people can overwhelm localities and their capacities of self-regulation, many of which have operated by strong cultural norms and shared values, the strong customs, high trust, and reputational controls of stable intergenerational communities, reputational forces, and other such things

Governments have long had to expand their scope to manage the transnational and international forces of a globalized economy and to enforce, police, and constrain interests of capital and aggregative systems of human action and communication that operate in vast networks that extend across the world and bring people under their sway. For instance, when AI companies can wield power far exceeding most states, something needs to be done to protect the common good from being destroyed by their private interests. This, again, is not an entirely new phenomenon. At its height, the East India Company’s private military was twice the size of the standing British Army. Unaccountable mercantile, industrial, technological, and digital forces are not unprecedented. The idea that the greatest threat is government can be very naïve to such realities.

If no action is taken, my suspicion is that AI and AI-like content will become pervasive in popular culture, with little to stop it, and that it will drive out most quality human content. Increasingly, entertainment produced for (and by) the general public will be largely AI-produced. Video and TV will be largely AI-produced. Music will be largely AI-produced. Advertisements will be largely AI-produced. Art will be largely AI-produced. Most of the text people encounter and produce will be AI-produced. Because of its derivative character, even the human creations upstream of AI will feel debased by AI’s constant semi-digested regurgitations of them. Because of the imitative character of creation, even the human creations of people surrounded by AI will themselves start to share its features.

Increasingly, services delivered to the public will be AI, with AI taking the place of and simulating what were once human interactions. Value and humanity will be drained out at every step. The ‘taste’ of many things may be the same, but the ingredients will have been completely altered. We will find ourselves living in the uncanny valley and most people will not register the significance of the disaster that has befallen us. It is already well underway, of course, and most do not care. AI could easily overwhelm the public education system, infecting every aspect of the learning process, demoralizing students and teachers. It will break talent pipelines. It will hasten a general move to a post-literate culture as people are denied the opportunity to gain a taste for something more substantial.

The quiddity of the world can never be destroyed. But as presented to us, it will be sapped as the artifice, the replication, the disembodied, the virtual, the spectacle, and the simulation take the place of the real and then remake it in their image. Most people, having been raised on spectacle and slop, will have little sense of its unhealthiness. An example of this is the way that real historic places can be reformed to look more like the hyperpalatable images that may once have been inspired by them. Medieval towns are refashioned to attract Instagram tourists raised on the artificial spectacles of Disney and Harry Potter, the confected image steadily consuming and effacing the original.

Now, again, as I have argued elsewhere, forms of ‘slop’ and other AI-like creations have long been with us, nor was this phenomenon restricted to low culture. We have lots of impersonal and engineered boilerplate language in terms and conditions and advertising emails, and other such documents, which, for their purpose, is often not inappropriate. However, we also have academic texts filled with undigested ideological jargon. We have work that is mindlessly derivative, bland, or sentimentalist, often cynically designed to spam or mash people’s emotional and aesthetic receptors. Much of the work of someone like Thomas Kinkade, for instance, feels like a sort of AI slop avant la lettre, which, while quite technically competent, is hyperpalatable, generic, uncanny, and insubstantial. We have websites filled with utterly formulaic articles, optimized to attract eyes—often through non-human algorithms—to generate ad revenue.

People producing such material have often complained about the constricting effect that such production has had upon them, stifling their humanity in service of the imperatives of efficiency, rationalization, monetization, and production. All these things have merely been waiting to be automated. The values were already in place; we were merely awaiting technologies to implement them at scale. So much of the perceived gains of the past couple of centuries have been a result of conforming ourselves to, rendering ourselves scrutable by, modeling ourselves with, identifying ourselves with the representations of, quantifying ourselves by, outsourcing our deliberation and agency onto, and submitting ourselves to manipulation by the Machine and its logics. Our inability to consider and weigh the costs that accompanied these ‘gains’ has much to do with the fact that they do not register within the Machine’s frames of value.

When people have long been writing for the Machine, it makes sense that they would want to offload that dehumanizing work onto AI, getting the Machine to write for them. So many of these activities are themselves results of a society that has already wholly submitted itself to a series of depersonalized, automated, standardized, and engineered imperatives and logics for the sake of maximization of productivity and exchange value. AI merely permits these imperatives and logics to be thoroughgoing, spreading their control to realms that once felt their dominion much less keenly.

For instance, universities have increasingly been perverted by the imperatives of the market, their internal formative ends being subordinated to external ends of revenue maximization, credentialism, careerism, and the like. The crisis of AI in the university is an amplification of the existing tensions that threaten the soul of the institution: if university is really about credentialism, careerism, and networking, why not cheat on your academic work? If university is merely about research generation, why not outsource this? If this can be undertaken more effectively with much diminished human involvement, why not adopt AI wherever we can?

Much of what we are concerned about relative to AI is the escalation of long-existing logics and trends and the trophic cascades that these can cause in society and its institutions. Society and its institutions require some degree of ecological balance and the uncontrolled implementation of a new set of technologies such as those named by AI can be catastrophic, collapsing whole ecosystems. Now that the generation of slop, mindless, and/or generic words can be automated on a mind-bogglingly immense scale, and with so much more ease than authentic words and creations, what was formerly an unpleasant yet more limited phenomenon threatens society in general with a Great Uncannying. Our spam filters fail and our societal inbox is filled with unwanted garbage. Rather than manifestations of human beings falling short of the vocations of speech and creation, we will be exposed to a myriad expressions of the non-human feigning humanity and of human beings outsourcing their human expression to the non-human.

When I shared some of the above thoughts a few days ago, someone responded by suggesting that we replace the term ‘TV’ for ‘AI’ and imagine ourselves back in 1955. What might we notice?

This, it seems to me, is a very worthwhile experiment.

Some people have a notion that those resisting the encroachments of AI are merely instinctively reacting against its novelty, perhaps animated by a constitutional dislike of change and undifferentiating prejudice against the new, or nostalgia for some fancied pre-technological idyll from which we have fallen.

It is helpful to recognize the strong continuities in technological criticism and how many of the strands of argument and consideration raised in response to AI are applicable to and have been developed in relation to precursor technologies, such as television. It is also important to appreciate that many of the people challenging many forms of AI adoption are wary of many very familiar technologies (such as television) and may even encourage more reflection upon and circumspection about our use of technologies that seem almost natural (such as the printed book). What many of us are concerned about in AI is not merely a saltational leap of evolution beyond our existing form of technocultural society, but an accelerated though incremental development of logics that are already endemic.

Such technological criticism is not the same as an aversion to any novel technology, nor does it treat a technology’s novelty as the primary criterion for our assessment of it. Some of our most formative technologies in the present are among the oldest technologies of all, things such as writing, the mirror, or money.

Likewise, technological criticism need not entail a resistance to using new technologies (AI among them). The accusation of being a ‘Luddite’, besides generally being wielded by people with limited sense of the Luddite movement, is inaccurate in the sense that it is usually intended. Our concern is that we use—or determine not to use—both new and familiar technologies in ways that are considered and discriminating, ordered by an understanding of such technologies, what they entail, and how they frame and form our reality. Such an understanding is not merely an understanding of a technology on its own terms—the sort of understanding typically pursued by early adopters—but an understanding of, among other things, the ecological societal effects it has.

It is interesting to revisit the criticism that responded to the screen age and to see how prescient much of it feels in relation to the Internet, social media, mobile devices, and AI. From Walter Benjamin’s ‘The Work of Art in the Age of Mechanical Reproduction’ and its recognition of the way that logics of creation and production transform our experience of things, to Guy DeBord’s The Society of the Spectacle and its exploration of how representation and the image consume and efface a reality that once stood over against them, drawing all into a sort of show-business, to Marshall McLuhan’s The Medium is the Massage, to Neil Postman’s Amusing Ourselves to Death with its discussion, among many other things, of the trivializing effect of television, to writings by Martin Heidegger, Jacques Ellul, Ivan Illich, Jean Baudrillard, and many others, it should be evident that AI is a further development and radicalization of technocultural logics that have been advancing dramatically over the past hundred years. The deeper logic and tendencies of these developments have long been recognized and presented as objects of concern and criticism. Critiques of AI are standing on the shoulders of several giants of twentieth century thought.

In our current contexts—with their growing tendency to fuse in a single common context of technologically-mediated and specious hyperreality—Susannah and I have spoken a lot about the need for something akin to the Arts and Crafts Movement, a movement that maintains and pursues excellence in the face of the plummeting of standards for the sake of mass output, while seeking to make beautiful, good, human, and humanizing creations available to everyone. We are also calling for a Christian humanism.

As Tyler Austin Harper argued in a recent article in The Atlantic, Christians have conceptual resources with which to speak to realities such as AI that non-Christians, even when they have a growing unease with AI, typically lack. Those who largely resist the move towards AI, yet do not merely retreat into elite enclaves, seeking rather to spread human creations to all, forming, defending, and extending communities of virtue, can really make a difference. And the Church is, as Christians have long recognized, an Ark amidst the Deluge. This is where our work must focus.

originally published at the Anchored Argosy