Ideology with Autocomplete: Reading David M. Berry on Synthetic Media and the Automation of Meaning

By J. Owen Matson, Ph.D.

Link to David Berry's Article in AI & Society

What Is Synthetic Media and Why Should We Care?

Synthetic media, were one to believe the current glut of industry white papers and keynote slides, is the latest miracle to emerge from the algorithmic sublime. We are told it is the dawn of a new epoch, a rupture in the history of representation, and—most urgently—a business opportunity. Yet David Berry, thankfully, resists the temptation to declare the death of meaning at the hands of deepfakes or to marvel, slack-jawed, at the sight of a synthetic pope in Balenciaga. His concern is not with spectacle but with infrastructure: not what synthetic media shows, but how it works, and what sorts of epistemic and political arrangements that work implies.

The paper opens with a clear distinction. Synthetic media is not simply another species of image or text, nor is it best understood through analogies with earlier representational forms. Unlike traditional media, which mediate the real through established aesthetic or ideological codes, synthetic media does not represent at all. It generates. It does not offer an interpretation of the world but produces outputs through stochastic sampling from training data. Berry is at pains to show that this is not merely a technical shift, but a philosophical and economic one. Synthetic media belongs to what he calls computational capitalism—a regime in which knowledge, culture, and perception are increasingly the result of model-to-model interactions, without the need for worldly referents or interpretive labor.

This is the wager of the paper, and it is a serious one. To treat synthetic media as an object of aesthetic criticism is to miss the point. What is at stake, Berry suggests, is not what synthetic media means, but how it recodes the very conditions under which meaning is made, trusted, and circulated. The terrain is not symbolic but infrastructural, and the politics is not about what media says, but how it trains, predicts, and composes. We are not witnessing a crisis of truth, but a reconfiguration of plausibility, governed less by narrative coherence than by probabilistic fluency. In this light, synthetic media is not a novelty—it is an epistemic transformation hiding in plain sight, one that demands a correspondingly materialist analysis.

Media No Longer Represents, It Recursively Predicts

Berry’s central claim is as provocative as it is lucid: synthetic media does not represent the world, it anticipates its own reception. It does not depict reality, it predicts it—though not in any epistemically noble sense. The referent, once a stubborn fixture of meaning-making, has here been quietly dismissed in favour of a recursive loop of plausibility. These systems do not ask “what is true?” but “what looks like something you might believe is true?” One might be forgiven for thinking that ideology has been outsourced to a kind of stochastic bureaucrat with a flair for mimicry.

The consequences of this are not merely ontological, but profoundly political. In the logic of synthetic media, meaning becomes a matter of probabilistic fluency—an endless recombination of patterns drawn from past data, filtered through the smoothing operations of machine learning. This is not media that speaks about the world. It is media that speaks from a model trained on the already-spoken. Or as Berry might have it, what we are witnessing is not the mediation of reality but the continual iteration of what reality was most frequently imagined to be. The model doesn’t so much lie as it hallucinates in statistically sanctioned ways.

There is something almost charmingly cynical about this. What is synthetic media, if not ideology with autocomplete? It supplies not beliefs, but the shape of belief. It is less a purveyor of ideas than a manager of expectations, tirelessly productive and almost entirely indifferent to the truth-value of its outputs. To critique synthetic media on grounds of “factuality” is to misrecognize its function. These systems are not failing to represent the real—they are operating precisely as designed, which is to say, circulating content whose primary virtue is that it resembles something you’ve already seen, liked, and scrolled past. The tragedy is not that they distort the world, but that we’ve trained them on our distortions.

Platform Capitalism and the Automation of Semblance

Berry situates synthetic media within the larger machinations of platform capitalism, which he portrays not as a neutral infrastructure but as the staging ground for a new mode of cultural production. Here, value is no longer primarily extracted from representation—as in the traditional industries of media or advertising—but from the automation of form itself. The platform no longer cares what is said, only that it conforms. What circulates is not meaning but its echo, packaged for seamless repetition. A cultural economy once fixated on the novel or expressive now finds its engine in the smooth reinvention of the statistically probable.

In this schema, the metric of success is not insight, but plausibility. Berry’s analysis highlights how synthetic media subtly but decisively dissolves the distinction between information, knowledge, and simulation. In a landscape governed by probabilistic text prediction, plausibility comes to replace truth as the governing epistemic value. This is not merely a matter of people being fooled—it is a matter of knowledge itself being redefined, not as justified belief, but as statistically probable coherence. The implications are not trivial. If the authority of knowledge rests not on epistemic warrant but on predictive fluency, then the very conditions under which sense is made begin to shift beneath our feet.

Platforms do not seek truths; they reward content that resembles what has already performed well. As Berry argues, plausibility has supplanted meaning not by accident but by design: it is easier to monetize patterns than propositions. One need not evoke Baudrillard to observe that the real is becoming less relevant than the recognizable. The simulation has not replaced the real; it has simply made it redundant.

There is no need to panic, though. Capitalism has always been rather partial to simulation. The commodity, as Marx noted, acquires its value not from its utility but from its capacity to signify—its aura, its brand, its symbolic function. Political speech, likewise, has long thrived on the production of resonance rather than coherence. Synthetic media may automate this logic, but it does not invent it. What it offers is a kind of hyper-fetishism: not the concealment of labour behind appearance, but the concealment of the absence of labour behind the endless recombination of prior appearances.

Berry is clear-eyed about this transition. His argument is neither alarmist nor utopian. He offers a sober analysis of how the automation of semblance allows platform capitalism to scale cultural production while divesting it of deliberation, conflict, and consequence. It is not that synthetic media cannot be meaningful; it is that under current conditions, meaning is beside the point.

Cognition Without Interpretation: The Epistemic Stakes

It is in his analysis of epistemology that Berry’s argument gathers its most unsettling force. The concern is not merely that synthetic media simulates meaning poorly, but that it increasingly displaces the interpretive labour by which meaning is made at all. The text, once a site of negotiation between expression and understanding, is now more often an artefact of computational recursion: a structure not addressed to a reader, but to a system trained to anticipate what readers are statistically inclined to receive.

Berry’s analysis echoes a key insight from N. Katherine Hayles: cognition is not the exclusive property of individual human minds, but a process distributed across material, technical, and social assemblages. The epistemic consequences of synthetic media, then, are not just about what humans believe, but about how meaning itself emerges through systems of recursive prediction and infrastructural design. He does not mourn the loss of the autonomous subject or suggest that cognition is vanishing from the earth. Rather, he reminds us that cognition has always been distributed—emergent from assemblages of humans and nonhumans, infrastructures and interfaces. The question, then, is not whether we are still thinking, but what kinds of thought are being scaffolded by these systems, and to what ends.

Synthetic media, in this account, does not so much think as it anticipates thinking. It constructs plausibility not through engagement with the world, but through recursive reference to its own proliferating output. When models train on the residue of their own predictions, the system begins to forget what it was once trying to learn. This is not a simple problem of quality control or hallucination. It is a form of epistemological auto-cannibalism, in which the conditions for sense-making erode not through negligence, but through perfect circularity. The medium learns to simulate cognition while increasingly detaching itself from the interpretive contexts that gave cognition its stakes.

Berry stops short of melodrama. He does not call for the abolition of synthetic media, nor does he lapse into sermonizing. But he does ask us to take seriously the erosion of friction, ambiguity, and resistance in environments where fluency becomes a proxy for thought. If there is a danger, it lies not in the existence of these systems, but in our unreflective integration into their logic. The risk is not that machines will learn to think, but that we will forget how.

Critical Critique without Humanist Nostalgia

Second only to the scope and nuance of his work, Berry’s strength lies in his refusal to sentimentalize the human. At no point does he resort to the tired pieties about irreducible creativity or the ineffable genius of the human mind. Instead, he takes seriously the proposition that synthetic media alters not simply what is produced, but the conditions under which production is intelligible at all. This is a rarer move than one might expect. While many critiques of generative AI become breathless in their defence of human authorship, Berry keeps his eye on the more fundamental question of how media systems structure epistemic possibility.

The essay deserves credit for maintaining a conceptual clarity often missing in critiques of computation. Rather than becoming entangled in abstractions about “post-truth” or “AI hallucination,” Berry anchors his argument in material infrastructures and their recursive logic. He makes visible the shift from reference to prediction not as a metaphysical puzzle, but as a function of systems optimised to simulate coherence. In doing so, he renders the strange perfectly legible—not through simplification, but through analytic patience.

This is, all told, a modest and elegant intervention. It does not shout, but it leaves its mark. In a discourse so often derailed by moral panic or technological determinism, such clarity is no small achievement.

Openings for Further Thought

While Berry is commendably free of both nostalgia for a golden age of unmediated truth and the sort of corporate enchantment that sees in synthetic media the dawn of a new Enlightenment, the essay might nevertheless have profited from a deeper entanglement with thinkers like N. Katherine Hayles, who have long insisted that cognition does not reside solely in grey matter or indeed in any single organism. Hayles’s insistence that cognition is the act of interpreting information in context, where context is itself mutable and distributed, provides a conceptual apparatus that complements the complexity of the systems Berry describes. One need not share all of her posthuman sympathies to see the advantage of treating meaning not as something housed in a sovereign subject, but as an emergent relation within a media ecology that has long ceased to flatter human exceptionalism.

This is less a criticism than a missed rendezvous. Berry speaks of critical literacy, a phrase that suggests both intellectual virtue and a BBC announcement on how to prepare for heavy fog. But he does not quite say what cognitive conditions would make such literacy possible amid the recursive churn of probabilistic semblance. The problem, one suspects, is not that we no longer know how to read, but that the act of reading has itself been redistributed, sampled, and indexed by systems whose output is not interpretation but pattern reinforcement. If the author has mapped the erosion of referential integrity, what remains is to ask how a different kind of interpretive agency might arise—not in spite of these systems, but in recursive relation to them. What is needed is not just awareness but apprenticeship in the ontological consequences of simulation. The task is not to reclaim our brains, but to understand how thought itself is being reformatted by the platforms that now host its semblance.

What this opens up, then, is not a lament for the demise of interpretation, but the possibility of reimagining it as a shared practice across uneven epistemic terrains. If Berry’s diagnosis tells us that semblance has usurped referentiality, then we might begin to ask what sorts of meaning can be made in a world where meaning no longer means quite what it used to. The challenge is not to restore the sanctity of interpretation as a solitary human act, but to situate it within cognitive assemblages that include predictive architectures, probabilistic grammars, and increasingly agential infrastructures. The old Enlightenment model of the reader as rational adjudicator gives way to something more dramaturgical, in which reading becomes a form of negotiation with the constraints and affordances of systems that do not understand what they generate. This is not the end of thinking, but the need for a new politics of cognition—less concerned with rescuing the autonomous subject than with tracing how agency flickers and reforms across circuits, prompts, and probabilistic phrasings. If Hayles invites us to see cognition as a process, Berry shows us what happens when that process is harvested. Together, they point toward a theory of reading appropriate to the age of synthetic media: recursive, distributed, and always slightly out of sync with itself.

The peril, in the end, is not that machines will outthink us, but that we will grow accustomed to thinking as they do. The real risk is less extermination than exoneration: relieved of interpretive labour, we may gradually forget that meaning was ever something to be wrestled with rather than accepted at face value. Berry’s essay reminds us that what synthetic media automates is not just language, but the very semblance of coherence, the illusion of thought’s completion before it has begun. To reclaim meaning-making, then, is not to clutch at some Romantic fantasy of deep human interiority, but to insist that even in an age of plausible text and algorithmic fluency, understanding remains a site of political struggle. In this context, literacy is not a skill but a stance—a refusal to let prediction pass for explanation, or fluency for depth. It is not that we must outwrite the machine, but that we must remember why writing mattered in the first place.