Beyond 'AI Slop': The Deeper Conversations We're Avoiding

Overview

The proliferation of low-quality, AI-generated content—colloquially dubbed "AI slop"—has triggered a wave of institutional pushback. This spring, major platforms and repositories, including arXiv, began banning submissions that bear the marks of unchecked artificial intelligence use. This move mirrors a broader trend across traditional media, social networks, advertising, and even legal fields, where synthetic text is flooding in at unprecedented scales.

While the presence of this content is undeniable, the linguistic framing of the problem may be obscuring the more significant societal debates. The term "AI slop" functions not merely as a complaint about a specific output but as a rhetorical device that allows users to smuggle in broader philosophical positions about the technology’s role in society.

The term “AI slop” hides conversations we need to have. When we call something “AI slop,” we’re often doing more than complaining about one output.

This analysis suggests that the discourse around slop lets us feel as though we are addressing important social issues without actually engaging in them. By examining the two dominant attitudes toward the term—quarantine and contagion—we can see how it serves as a slippery vehicle for avoiding deeper discussions about AI’s proper place in our culture.

The Two Attitudes: Quarantine vs. Contagion

When critics and proponents alike dismiss AI outputs as "slop," they are rarely just describing the quality of a single piece of generated text. Instead, they are invoking one of two broader worldviews regarding the technology’s future and utility. Understanding these distinct attitudes is crucial for decoding why the term has become so polarizing.

The Quarantine Attitude

For optimists, the quarantine attitude views AI slop as an unfortunate anomaly rather than a representative feature. In this framework, calling an output "slop" is an attempt to cordon off failures from the genuine value AI can provide. Proponents argue that useful applications—such as accurate translations, helpful code generation, or new mathematical discoveries—are distinct from the degenerate cases produced by reinforcement learning optimization.

This perspective was evident when OpenAI launched Sora 2. Sam Altman framed creativity as heading toward a "Cambrian explosion," while characterizing the RL-optimized slop feed as a degenerate case that the company had taken great care to avoid. Similarly, Google attempted to quarantine embarrassment when its AI search suggested using glue to keep cheese on pizza. The company attributed such results to uncommon queries, effectively sealing off the embarrassing case to ensure the broader rollout could proceed uninterrupted.

The Contagion Attitude

In stark contrast, the contagion attitude posits that slop is not a defective specimen but a revelation of what AI truly is. Critics holding this view argue that the technology is inherently derivative, careless, and hostile to the culture it feeds upon. To them, the flood of synthetic children’s books, unwritten student essays, and hollow apology emails are not unfortunate by-products but perfect representatives of AI’s capabilities.

Philosophers like Madsen and Puyt have argued that the flood of recycled output is not a defect to be cleaned up but rather the essence of what AI amounts to. This sentiment was echoed by Charlie Warzel in The Atlantic, who responded to Sora 2’s launch by declaring that "AI slop is winning." He characterized Altman’s vision not as an explosion of creativity, but as "a world of creativity without craft."

The Rhetorical Slip

The danger lies not in either attitude individually, but in how the term "AI slop" allows speakers to slide between them without explicit declaration. The term presents itself as a neutral verdict on an output while implicitly suggesting a verdict on the entire technology. This linguistic sleight of hand is similar to the common critique that a text merely "sounds like AI."

Thick Terms and Disgust

Philosophers of language classify "AI slop" as a thick term—a word that describes and evaluates simultaneously. It invites a visceral reaction of disgust, prompting audiences to recoil. However, this recoil carries different implications depending on the speaker’s intent.

• In the optimist’s mouth, disgust is a push to quarantine the bad output from AI’s valuable work. The emphasis falls on the "slop."
• In the critic’s mouth, disgust is a push to scuttle the entire enterprise. The emphasis falls on the "AI."

Because the term operates on two levels, two people on opposite sides of a major social debate can walk away from a conversation thinking they agree. They have settled that some output is AI slop, but the larger disagreement about the technology’s value remains hidden and undefended.

The Historical Precedent of Slop

This rhetorical slide can be exposed by recognizing that slop existed long before artificial intelligence. There were always bad novels, formulaic films, empty lectures, and academic committee reports that said nothing at length. No one concluded that bad novels settled the worth of literature or that worthless reports invalidated the existence of academic committees.

The same logic should apply to AI. The slop emanating from our machines does not prove that AI is worthless, just as its successful outputs do not prove it is wholly good. When discussions of "AI slop" seem to bear on the overall worth of the technology, it is typically a background view about AI’s proper role that is doing the heavy lifting. This underlying view is carried in quietly and often left unexamined.

The Need for Explicit Discourse

The current reliance on "AI slop" as a catch-all complaint prevents society from having the necessary conversations about regulation, ethics, and integration. If we are to move forward productively, we must separate the aesthetic judgment of an output from the structural questions about the technology.

Identifying the Real Debate

The core issue is that the term allows us to bypass the hard questions. Instead of asking whether AI-generated content should be regulated in legal briefs or whether synthetic media distorts educational standards, we simply sneer at the quality. This avoidance is dangerous because it leaves the background assumptions about AI’s role in society open to default enforcement rather than democratic deliberation.

To address this, we must demand clarity in our discourse. We need to explicitly state whether we are criticizing the quality of a specific generation or the systemic incentives that drive low-value production. Only by separating the "slop" from the "AI" can we engage in the substantive policy and ethical debates that the technology demands.

Moving Forward

Both attitudes toward slop have merit. The quarantine attitude rightly reminds us that not every AI system or output is equivalent. The contagion attitude correctly warns us to remain vigilant against incentives that reward cheap, low-value production. However, the term itself has become a barrier to clear thinking.

We must resist the temptation to use "AI slop" as a shorthand for disapproval that smuggles in uncontested philosophical claims. By unpacking the term, we can finally address the real questions: What role should AI play in creative and intellectual endeavors? How do we distinguish between useful augmentation and derivative replacement? And what structures do we need to ensure that AI serves human interests rather than merely optimizing for engagement?