Critical Essay: Simulated Consensus

Critical Essay: Simulated Consensus

A Hobbes Machine?: Ryuta Aoki’s Simulated Consensus

by Seo Dong-Jin, Professor of Intermedia Art, Kaywon University of Art & Design.

Originally published in the ACC CREATORS Residency 2024 catalogue. Reproduced here courtesy of National Asia Culture Center (ACC) and Seo Dong-Jin.

Ryuta Aoki’s work Simulated Consensus is exceptional in the way it deviates from the typical approach found in the visual arts, where a sense of enthusiasm and fascination over artificial intelligence (AI) prevails. It does not align itself with any crude paeans to the economic impact of AI, nor does it limit itself to attempts to expand beyond tedious and fruitless philosophical discussions on the relationship between the real and virtual and questions about the “reality” of what AI produces. Instead, Aoki poses questions about the potential of AI as a political machine. In that sense, his work occupies a special place in ACC CREATORS Residency 2024 AI·Human·Multiverse. Most crucially, his work goes beyond others that focus on the visual illusions that AI is capable of producing, as it concentrates instead on its role in terms of political perception.
Simulated Consensus interrogates the idea of “voting” as the most representative symbolic act of the democratic process. A liberal representative democracy simulates the creation of a “general will” of society. According to this, the individual expresses their own will through the act of voting, with the general will emerging as the sum total of these personal choices. Based on this, the artist uses the term “consensus” to refer to what is “calculable” as an average of values obtained by manipulating and limiting different variables. He focuses on the point that “public opinion” as an effect of these calculations constitutes the realization of the political representation presumed by liberalism. It is a vast calculating machine, but it is not a mere counter adding up ballots. It involves voting through a large language model (LLM), and the device through which voting takes place is one that produces AI-generated voter personae. Aoki is imagining the potential for an AI-modeled voter-citizen.
In that sense, his questions take on multiple layers. To begin with, as he foresees the potential for AI to operate as a political actor, he asks what effects it might produce within the “device” of democracy. If AI-generated political agents can be regarded as subjects making the same choices as human ones, it may then be possible to transplant into reality a world of linguistic models oriented toward specific choices, and to thereby manipulate political choices. Obviously, this is not at all unfamiliar to us. Various phenomena—the endless controversy surrounding the manipulation of online posts, internal platforms that revolve intensely around specific keywords and thumbnails, and the algorithm-driven amplification of public opinion on social media, which some have described as an “outrage machine”[1]—all allude to the ways in which “consensus” has been simulated and taken the place of democratic representation. As a result, acts of representation and choice may end up dictated by the “opinion” that encourages, promotes, and amplifies them, rather than a process of independent will and determination. South Koreans recently experienced a tragic outcome of this with a fanatical political leader who declared martial law based on the illusory world (simulated consensus) portrayed in YouTube videos by paranoid far-right political groups. While his insurrection attempt may have been thwarted, it certainly gave a renewed awareness of the ways in which algorithm-driven YouTube videos have ended up creating a linguistic world of paranoid fantasy and representations of reality.
There is still another layer to the questions about democracy’s a priori assumptions in Aoki’s project. These are also matters that relate to a fundamental dilemma of liberal democracy. They concern the premise of liberalism viewing the functioning of reality as an outcome of personal choices. These questions cannot be separated from the irony of today’s algorithm-driven collective acts. Signs of the crisis in class politics can be detected in the behaviors of workers who cast their votes for far-right authoritarian politicians who oppose the voters’ own class-based interests, as well as left-wing voters who take political correctness to extremes as they portray the key social forces capable of realizing their political visions as shallow, narrow-minded figures worthy of mockery. This conception of political ethics limits the establishment of social life (class struggle) as a theme in politics. The liberal economic order that has emerged in the wake of neoliberal globalization is one that has resulted in the kind of extreme inequalities that make terms like “polarization” seem inadequate. Financialized economies have turned jobless growth into a routine phenomenon. Unstable working conditions are widespread, as exemplified by the phenomenon of “platform labor”; the rights attendant on labor (what has typically been referred to as “social security”) are being stripped away. Yet the state of what some have described as “work without workers” has foreclosed the possibility of individuals organizing as collective social agents—as a working class, in other words—and subjectivizing their own experiences, consciousness, and attitudes[2].
In that sense, the politicization of AI is all the more threatening. It reduces collective interests to a calculation of individual choices. Moreover, those personal choices are presented from moment to moment as “opinions” or “consensus” that have been selected, categorized, and averaged by AI. The potential for us to reflect on our accumulated experience and form our self-awareness on that basis has been taken away, leaving us the sole option of following along every time. To quote the words of Ryuta Aoki, we are automatically transported into a world of “simulated consensus.” It goes without saying that the world in question is the product of a behemoth information and communication industry representing the combination of big tech with vast business interests. It is a framework that we are obliged to redesign into something that stands up to huge platforms that would privatize collective thought and experience. In many countries where platforms have already established themselves as their own social infrastructure, the risks posed by the information and communications industry are not limited to the profits that are claimed through platforms; they endanger democracy itself.
In this regard, it is false to conclude that populism is responsible for threatening liberal democracy, or that social media and other platforms are the forces lurking behind that. We are fooling ourselves if we believe that populism represents the opposite of an innocent liberal democracy. When the representation of social interests is frustrated, populism seizes the chance to do so. This manifests in forms such as people choosing a charismatic leader (effectively the same choice that an individualized consumer makes on the market when deciding which brand of product to purchase) or following along with trends of opinion that center on inflammatory viral posts (much like the choice to participate in a trend or fashion that is driving the market). They are constantly drifting in search of a fantasy figure who promises to realize their thwarted desires. This means that until we redesign the hypotheses of a liberal democracy system that is programmed by AI democracy, we will never escape the world of simulated consensus. This is not utterly out of the question. When Salvador Allende began his journey toward democratic socialism in Chile, he attempted to design a form of “AI” that could achieve autonomous rule for the masses. We should not forget that when the US-backed general Augusto Pinochet carried out his bloody coup d’état, the dream he crushed was that of a grass-roots AI—a vision that rested in the utopian drive to achieve a new society[3].
In Ryuta Aoki’s Simulated Consensus, the idea of AI as a Rousseauian machine of politics/communication—a democracy machine manufacturing a general will—is something quite fluid in nature. It rests upon AI as a vast machine within a platform that firmly supports a world where the idea of the “end of history” prevails: the governing democracy of today where people believe there is no alternative to liberal democracy. Yet his project also alludes faintly to the possibility of recalibration through the workings of a new form of independently designed AI. Aoki’s curious history only adds to the anticipation. We should not lose sight of the fact that he is an engineer who previously worked at a computer communication company. He also holds the idea that he is both an artist and a social entrepreneur. This may explain why he refers to his work as “social sculpture.” In other words, he envisions artistic practice as the designing of lives in society. In that sense, it may be inevitable that he would respond so keenly to the hideous forms of sociality that AI creates. Democracy is a political institution and mechanism that produces sociality under the names of “consensus” and the “general will.” Yet there are competing democratic models within it. Simulated Consensus is a project that experiments with warnings of the potential for bias in AI, but it is also a critical point of departure that encourages us to realize the model for democracy contained within AI. I look forward to seeing the next stage of updates in Aoki’s version of AI democracy.
 
[1] Tobias Rose-Stockwell, Outrage Machine: How Tech Amplifies Discontent, Disrupts Democracy―And What We Can Do About It, Korean translated by Hong Seonyeong (Sigongsa, 2024).
[2] Phil Jones, Work without the Worker: Labour in the Age of Platform Capitalism, Korean translated by Kim Gomyeong. (Rollercoaster, 2022).
[3] Seo Dongjin, “An Incoherent Techno-Utopia: The Ideology of the Fourth Industrial Revolution,” The Quarterly Changbi, vol. 45, no. 3 (2017).