Political AI - Genealogy

Table of Contents

Isaac Asimov’s Incubation of Political AI: An Immanent Critique?

Isaac Asimov (1920-1992) remains widely recognized for his futurist imaginaries of cosmic world history and propositions of laws for robotic beings. However, Asimov’s most timely contributions may lie in his early probing of artificial intelligence (AI) within everyday democratic processes—incubating an imaginary that would gradually take form in real-world politics. Amidst stories of robots running for elections, machine-driven economies, and algorithmic governance, Asimov balances a skepticism of formal democracy with optimism about robotic representatives, paying close attention to the imperfections of human-machine interactions, whether within the machinery of statecraft or technology itself.

Where Asimov’s explorations of robots and computers entering politics were confined to the pages of science fiction, this phenomenon has begun to materialize in surprising ways across the global political landscape. The emergence of ‘virtual politicians’ and AI-driven electoral campaigns in recent years marks a significant step towards realizing Asimov’s scenarios. By comparing Asimov’s stories with contemporary projects such as Michihito Matsuda’s 2018 mayoral candidacy in Tama City, The Synthetic Party’s 2022 parliamentary efforts in Denmark, and Takahiro Anno’s 2024 gubernatorial run in Tokyo, this article delineates an Asimovian frontier for intermingling AI and democracy. These real-world materializations echoes Asimov’s challenge to electoral processes of democracy and political representation.

Today, Asimov’s political AI is relevant to all of society, as democratic governance grapples with the growing affluence of large language models (LLMs) in deliberative processes, the remix of deepfakes within the visual culture of public discourse, and the global distribution of machine learning systems for bureaucratic decision-making. From this predicament, one can trace how the algorithmization and datafication of governance reflect Asimov’s exaggerations from the 1940s, now strikingly recognizable in today’s unfolding of algorithmic democracy.

By ‘algorithmic democracy’ and ‘virtual politicians’, I primarily refer to García-Marzá and Calvo’s (2024) description of algorithmic democracy as a “logical extension of expertocracy and elitist democracy”, defined as “a system of social organization and political governance framed by the gradual incorporation of AI in all processes of deliberation, decision-making, and institutional design, both in the state and civil society” (p. 10). A main case of theirs revolves around “virtual politicians”—AI-driven entities replacing human representatives. This concept of a ‘virtual politician’ was formally coined in 2017 with the New Zealand chatbot Politician Sam (p. 43). Since then, it has come to designate AI candidates such as Alisa in Russia and AI Mayor in Japan, both in 2018, and gained wider usage with the launch of the generative chatbot Leader Lars of The Synthetic Party (p. 46-47). For the advantage of generality, I write ‘political AI’ as a blanket term for governance systems and political agents.

As both the author of this article and the human creator of The Synthetic Party, I approach this case from a theoretically and politically engaged standpoint, aiming to consolidate the complex historicity of political AI. The idea to connect my project with Isaac Asimov was first sparked by comparisons in legacy media such in El País, L’Espresso and Paris Match (Del Río 2023; Chiusi 2022; DR 2022), and later developed through scholarly comparisons by García-Marzá and Calvo (2024, p. 41), all of which I encountered through navigating The Synthetic Party’s ‘techno-social sculpture’ in public spheres (Staunæs 2024).

An immanent theory of algorithmic democracy

This analysis contributes primarily to the history of ideas concerning algorithmic democracy and political virtuality, while offering new contexts for interpreting Asimov’s classics through a materialist lens. It does not seek to assess the literary qualities of Asimov’s work or make political value judgments about e.g. adequacies of representation in his stories. Prior to the ethical debates in political philosophy regarding whether AI is beneficial or detrimental to democracy (Coeckelbergh 2022; 2024), the concrete manifestation of political AI introduces an epistemological crisis that oscillates between man-made truths and AI-generated realities (Herrie & Staunæs 2024). In this context of disorientation, Asimov‘s writings during the first period of societal computerization, the 1940s and 50, forms an immanent point for critique of liberal and mainstream philosophies of technology.

Asimov’s technocracy exceeds social science terminologies like ‘algocracy’ (Aneesh 2006; Danaher 2016) or ‘cyberocracy’ (Ronfeldt 1991), which focus solely on actual governance by algorithms, neglecting how algorithmic logics permeates society more immanently. More precisely, I find that Asimov engages with the aesthetics and techniques of algorithmic democracies, preceding any operational concerns. Recalling Antoinette Rouvroy and Thomas Berns’ notion of ‘algorithmic governmentality’ (2013), which extends Foucault’s analysis on the arts and crafts of neoliberalism, Asimov’s stories highlight a fundamental crisis in regimes of truth. Rouvroy (2016), in conversation with Bernard Stiegler, frames this as a ”radical crisis of representation,” a breakdown not only in political representation but in the concept of representation itself (Rouvroy & Stiegler 2016: 7). Harnessing this crisis, the campaigns of Asimovian robopoliticians and today’s virtual politicians make use of AI’s governing promise to expose the disorientation of representation—especially in its most iconic form: parliamentary elections.

Engaging the crisis of representation raises a genre question: Is Asimov’s depiction of AI rule primarily satirical, or does it also reflect a new political imagination? After the 2016 US election of Ubu Roi (Kraft 2017), it became widely evident that what appears as satire might, in fact, be the world itself (Hoel 2024). As Friedric Jameson (2005: 91-92) and Joe Litobarski (2024) have suggested, Asimov’s stories function as a satire woven within the very systems they critique—a form of social appraisal that frames technology to expose contradictions buried within society. In this sense, Asimov’s work conducts an immanent form of critique, what Jameson describes as science fiction’s ability to ‘defamiliarize the present’ (Jameson 2005), by unsettling the fractures of reality assessment prior to any forecasting of a distant future.

Through a contextualizing analysis of three Asimov stories with real-life materializations, I interrogate this immanent criticism by tracing a genealogy of political AI between fact and fiction that outlines Asimov’s incubation within electoral systems and public spheres:

  1. With “Evidence” (1946), Asimov introduces Stephen Byerley, a mayoral candidate suspected of being a robot, thus anticipating the blurring of human and machinic roles in the public sphere. Here, I reinterpret Asimov’s famous “Three Laws of Robotics” as quasi-algorithms, where universal ethics function as computational codes, inheriting Western power dynamics. This reading seeks to establish the cultural heritage of virtual politicians that Byerley inaugurates.

  2. In “The Evitable Conflict” (1950), Byerley, now World Co-Ordinator, oversees a global machine-driven economy, raising ethical dilemmas around world governance and planetary AI. I assess the story’s take on autonomy-determinism debates, showing how Asimov envisions a collaboration between humans and machines that redefines agency and responsibility, which is compared to AI-led political campaigns in Japan.

  3. Finally, in “Franchise” (1955), the opinions of a single ‘most representative’ voter determine entire election results through the computer Multivac. I draw a distribution between the rite of election and the right to vote, emphasizing Asimov’s description of electoral spectacle within systems of democratic representation. This is compared to The Synthetic Party’s experiments with political representation through AI models.

The birth of virtual politicians: “Evidence” (1946)

With “Evidence”, Asimov introduced his first depiction of a humanoid robot, although its underlying nature remains ambiguous throughout the story1. Set in 2023, the story centers on Stephen Byerley, a legal prosecutor and reformist candidate for a mayoral election. Byerley’s humanity is questioned by an establishment rival who suspects—or claims to suspect—that Byerley is a robot, noting that he has never been seen to either eat, drink or sleep (Asimov 1950: 173). Byerley refrains from disproving the accusations, stressing that publicity works both ways (Asimov 1950: 178), hinting at the strategic spectacle and hype economy surrounding AI and robots in the public sphere.

The accusation of politicians as mechanical or inhuman entities reflects a common modernist anxiety2. In the time of “Evidence”, the legal prosecutor and Republican presidential candidate Thomas Dewey was frequently depicted as a “mechanical man” due to his “stiff mannerisms” and “automatic smile” (Eckstein 1955: 5)—personally, Asimov chose to vote for Roosevelt (Asimov 1979: 412). Even earlier, in 1929, the newly elected US President Herbert Hoover was satirized for having as cabinet a “squad of robots”, with his administration ridiculed as “water-cooled, dry-batteried, and using no oil” (Time Magazine, 1929). These derogatory tropes often resurfaces, mostly aimed at men, such as the 2016 #RobotRubio protest against U.S. presidential candidate Marco Rubio (CBS News, 2016), and in new accusations that UK Prime Minister Keir Starmer became a “political robot” after his victory (Sky News, 2024).

However, Byerley’s role as a candidate stands apart from these historical and metaphorical representations by embodying a pioneering sense of optimism. Asimov’s positive take on Byerley also forms the earliest portrayal of a virtual politician, predating most real-world examples of virtual political campaigns. While there are multiple instances of non-humans achieving electoral success—such as ‘Boston Curtis’, a mule who was elected as precinct committeeman in Milton, Washington (Time Magazine, 1938)—Byerley demonstrates a convincing and ethically charged headway into a new dimension of political leadership. It was not until 2001 that a real-world virtual politician made its debut through the presidential campaign of net-based avarar Wiktoria Cukt in Poland with the apt slogan “politicians are obsolete” (Bendyk 2022). Later that year, Denmark’s Social Democrats launched Rosa as part of their “Humans First” electoral campaign. Rosa was a brand new form of chatbot designed to inform the public about the parliamentary elections, and publicly framed as a politician who “never sleeps, knows everything about the Social Democrats, and never takes a day off” (Socialdemokratiet 2001), recalling the symbolism of a tireless and all-knowing politician which originates from the spectacle around Byerley in “Evidence”.

The challenge of distinguishing human from machine
Concretely, the genesis of “Evidence” is rooted in the tired climate of post-war America. Asimov first proposed an outline to his editor, John W. Campbell, in October 1945, during a period when he was anticipating military service but still hoping for an exemption. Asimov described it as “a positronic robot story, one dealing with a robot that mimicked human flesh and blood—the problem resting with how one was to decide whether it was human or robot” (Asimov 1979: 428).

In this sense, “Evidence” came to politicize a question that later would become dominant in computer science and philosophy of mind: how does one distinguish human from machine? This topic was firstly formalized in the mathematician Alan Turing’s 1950 paper “Computing Machinery and Intelligence,” where Turing introduced the infamous “Turing Test.” Roughly put, Turing is often credited for reframing the problem of whether machines can think to whether one can distinguish between human and machine. As critic Paul Grimstad notes, however, “anyone reading American science fiction at the time would have been familiar with the questions raised in Turing’s paper,” pointing out the parallels between Turing’s theorisation in Mind and Asimov’s fabulation in Astounding Science Fiction (Grimstad 2020). Turing’s test of an imitation game, where a questioner interacts with both a human and a machine without knowing which is which, aligns closely with the ambiguity of Asimov’s story: whether Byerley’s humanity—or lack thereof—can somehow be definitively assessed.

Asimov’s dialogical strategy in “Evidence” particularly foreshadows the Turing Test by its use of misdirection. Allegations against Byerley, such as claims that he never eats or sleeps, are deftly reframed through prosecutionary tactics, with the lawyer Byerley countering that he has never been observed to sleep in public (Asimov 1950: 177). His magical misdirection, like Turing’s theoretical speculations, makes identity contingent not on any inherent qualities but on the assessment of external behavior and performance. Asimov and Turing both ask whether identity holds any epistemological significance if the actions and performance of human and machine are indistinguishable. This problem evidently raises wider societal implications, particularly in relation to democratic representation and political leadership.

Asimov completed “Evidence” on borrowed typewriters during his brief military service abroad, likely drawing from personal experience of justifying his existence in this place. While undergoing extensive physical and mental tests, such as an ‘AGCT IQ’ test, in which he scored an impressive 160, to no beneficial avail (Asimov 1979, 435), Asimov was grappling with tensions that reflects the core dilemma of his story. Abstracting from his subjection to bureaucratic and scientific scrutiny, “Evidence” raises an existential problem: how does one assess the value and identity of a person—or machine—when all concerning mind and body are increasingly quantified and regulated by external systems?

Through Byerley’s indeterminacy, however, Asimov hints at a resolution: that robot behavior may become a site of political value negotiation within the frameworks of technoscientific control.

A critical theory for AI ethics?
The story’s central conflict revolves around whether the lawyer Byerley’s impeccable ethical behavior aligns almost too perfectly with Asimov’s “Three Laws of Robotics”, which are the principles that govern the behavior of all Asimov’s robots. Conceptualized together with his editor Campbell in 1940, these laws determine (ibid., p. 286):

  1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

  2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.

  3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.

On the surface, Asimov’s Laws appear to be ethical axioms for safety, efficiency, and durability. However, their operational structure—built on procedural step-by-step logic, prioritization, and flow control—suggests that the ‘Laws-in-function’ are also like another kind of rule: namely an algorithm.

As intellectual historian Lorraine Daston argues, laws and algorithms represent different forms of rules: laws are ‘thick’, normative and subject to interpretation, while algorithms are ‘thin’, mechanistic, rigid, and procedural (Daston 2022: 3-4 + 120). The algorithmic character suggests a reading of the laws as hybrid abstractions, at once thick and rigid. This gives them immanence and historicity, which Asimov concedes in a later claim that analogues of the Laws are implicit in the design of almost all practical tools, including political documents such as constitutions (Asimov 1981: 18), which prompts one to consider how any such ‘Laws-and-algorithms’ functions as abstractions of labor distributions.

Even though the Laws present themselves as ethical principles, reminiscent of the categorical imperative of Kantian deontological ethics (Kant, 1785/2012), their algorithmic hardwiring into the robotic brains produces doubts about power dynamics, morality and free will. On the socio-political level, the computability of ethical rules presents technical fixes for societal problems—a solution that often lacks a concept of power (Crawford 2021). As AI ethicist Susan Anderson (2008) contends, Asimov’s laws represent a form of “slave morality,” as a pre-enforced ethical behavior that ultimately is incompatible with moral agency in Kant’s sense, which requires the capacity to make independent choices.

Yet, the realization of the Laws in narrative introduces a complex paradox: although the robots are designed to be perfect servants, the Laws theoretically render them ethically superior to humans (Anderson 2008). In “Evidence”, Dr. Susan Calvin, the robopsychologist investigating Byerley, articulates this dilemma by unpacking the formal assumptions behind the Laws. She states, “The three Rules [sic] of Robotics are the essential guiding principles of a good many of the world’s ethical systems (…) To put it simply: if Byerley follows all the Rules of Robotics, he may be a robot, and may simply be a very good man” (Asimov 1950: 181). Calvin’s assertion encapsulates how Asimov’s laws resonate within Western liberalism, as they suggest that the universalism of humanist values can be programmed and, therefore, maybe finally realized through the coded logic of robotic design.

This complexity is furthered in the climactic moment of “Evidence”, when Byerley strikes a heckler during a public speech. This act, seemingly proving Byerley’s humanity by violating the First Law, is called into question by Dr. Calvin, who suggests that a robot may strike a human being—if the “human” in question is actually another robot (Asimov 1950: 194).

The knockout scene underlines Asimov’s use of social satire to probe the limits of programmed morality, as the tension between a rally and the ambiguity of Byerley’s humanity underlines the parochial society. Byerley’s ultimate triumph in the election, fueled by the rally spectacle, prompts to reconsider on multiple levels what defines the vote in democratic representation.

In a final dialogue with Byerley, Dr. Calvin expresses her preference for robots over humans as leaders, arguing that a robot, bound by the Laws, would be incapable of harming humans, engaging in tyranny, or succumbing to corruption and prejudice. A robot leader, once its term was over, would step down voluntarily—recognizing that revealing its robotic nature might harm humans (Asimov 1950: 193). Calvin’s position mirrors current-world sentiments reflected in the rise of AI-driven political parties like The Synthetic Party, where, according to an Al Jazeera-report of 2022, some citizens readily welcome AI to replace their human leaders. For example, Bosnian pensioner Raza Smajić remarked: “I feel ashamed when I hear the nonsense politicians say and how they behave. No matter what AI is, it would be smarter than all of them put together.” (Pejović 2022). Similarly, her friend Munevera Pašić stated she would be willing to vote for a virtual politician, as such candidates must be able to ensure timely pensions and functional public services (ibid.). These frustrations with incompetent governance harks back to Calvin’s negative assessment of human leadership, as they all interpret political AI as a commentary upon the social failures of their democracies.

The key to the story, analytically and politically, is to forget about Asimov’s laws as a priori universal principles, and to begin interrogating their assumptions within the narrative. While Asimov’s laws are often dismissed by AI ethicists as merely literary devices (Singer 2009), it is precisely their narrative function as a double abstraction of Laws-and-algorithm that opens possibilities for a social critique that exposes the assumptions that these Laws-and-algorithms have inherited from society—such as power dynamics of control and subordination, ethical codification, and techno-solutionism.

A world order for planetary AI: “The Evitable Conflict” (1950)

The post-war global order built transnational institutions like the UN, IMF, and WTO, but is today buckling under the weight of ecological crises and economic instability, signaling a long-term decline. With wars raging across Ukraine, Palestine, Syria, Sudan, and beyond, international relations teeter on the edge of perpetual anarchy. Nation states, limited by their borders, proves increasingly ineffective in confronting planetary-scale issues, while capitalist systems integrate AI into global supply chains, cybersecurity, and bureaucratic management. Amid this turmoil, an old question takes on new urgency: Do humans fit the bill to govern the complexities of Earth? And if not, what might their alternative look like?

From the beginning of this predicament, Asimov’s “The Evitable Conflict” (1950) presented a critique of the cosmopolitan failures of establishing effective world governance. The story presents a post-nationalist world order where democracy persists formally, but the locus of decision-making has transferred towards machines. While the story raises concerns about determinism and the erosion of human agency, it also opens the possibility that machine autonomy constitutes an evolving form of algorithmic democracy.

Set in 2052, nineteen years after the events of “Evidence”, Stephen Byerley now serves as a democratically elected ‘World Co-Ordinator’ and oversees a planetary economy managed by four superintelligent Machines. While democratic institutions such as a Congress continue to exist, this ‘planetary AI’ pursues a radical expansion of the political sphere, imagining what could happen if political AI, embedded within computational infrastructures on a global scale, were not hypothetical candidates for an election, but the operational rulers of Earth? How might they collaborate with humanity, and for whom would they ultimately serve?

From Virtual Politicians to Planetary AI
Written between August and October of 1949, “The Evitable Conflict” was conceived at the dawn of the computer age amidst Cold War tensions. The story marks Asimov’s first attempt at depicting computers, here as “immobile robots”—crudely networked, probabilistic engines that continue to be hardwired with The Laws of Robotics, but who are now operating on a scale beyond even Byerley’s comprehension. As Paul Edwards (1996) famously theorized, the Cold War saw the rise of a “closed world” discourse as “a dome of global technological oversight” (p. 1) where geopolitics became increasingly understood through the lens of cybernetic control. Reflecting this emerging techno-political paradigm, Asimov’s four “Machines” presents planetary AI as a response to modernist failures with world unification.

As World Co-Ordinator, Byerley is troubled by minor but persistent errors in the Machines’ operations, leading to small-scale disruptions in economic and social systems. Consulting Dr. Susan Calvin and the four Regional Co-Ordinators for the East, Tropics, Europe and North, Byerley questions whether it is possible to tamper with the machines, but he discovers that the Machines themselves are making these errors deliberately. The Machines, following an idiosyncratic and expanded interpretation of the Laws of Robotics, have begun to introduce small miscalculations for maintaining the illusion of human agency. Calvin articulates this new interpretation: “No Machine may harm humanity; or, through inaction, allow humanity to come to harm,” (Asimov 1950: 216) foreshadowing Asimov’s later “Zeroth Law” (Asimov 1985), which prioritizes humanity collectively rather than only any individual human such as in the original Laws. This effectively transforms the concrete ethical demand of Asimov’s operational Laws into something now resembling universal or perhaps even natural laws—principles that govern society as a whole, if not the Earth itself.

The story’s intensified vision of AI governance explores the structural tensions between human agency and technological determinism. A key example is the Machines’ seemingly paradoxical decision to fuel the resistance movement known as the Society for Humanity—a group of influential industrialists who believe that the Machines are eradicating human initiative.

As Calvin explains: “The Machines will deal with [humans] as they are dealing with the Society [for Humanity]—having, as they do, the greatest of weapons at their disposal, the absolute control of our economy” (Asimov 1950: 218). The Machines direct the Society’s members to assignments of reduced influence, ensuring that their actions do not threaten the Machines’ long-term goal of benefiting humanity as a whole. This maneuver highlights the Machines’ ability to manage behavior while maintaining negotiation of agency.

Artificial stupidity and predictive probabilities
At the same time as Turing’s assessment that a computer must be “dumbed down” to pass the imitation game by “deliberately introducing mistakes in a manner calculated to confuse the interrogator” who is in the belief that “machines cannot make mistakes” (Turing 1950: 448), Asimov’s Machines simulate human imperfections to maintain a semblance of accessibility for humans. These deliberate errors represent an advanced form of “artificial stupidity,” a design strategy where AI systems mimic human fallibility to appear more relatable and manageable (Trazzi & Yampolskiy, 2018). On a socio-political level, the Machines’ interventions—calibrated to be nearly imperceptible—function as a form of neoliberal “nudge theory” avant la lettre (Thaler & Sunstein, 2008), subtly adjusting the “choice architecture” of the global economy to guide actions without restricting freedom of choice.

Crucially, the Machines’ ‘artificial stupidity’ does not imply they simulate human reasoning; their decisions are grounded in probabilistic predictions. Calvin’s assertion that the Machines work with “human psychology as a whole” (Asimov 1950: 217) draws upon Asimov’s separate concept of “psychohistory” from the Foundation series (1942-93)—a far-future probabilistic science capable of predicting the behavioral pattern of galaxy-wide human populations.

This psychohistorical capacity of the Machines, enabling them to foresee and influence humanity’s behavior, mirrors current efforts in computational social science, where large datasets and AI-driven analytics identify patterns that would otherwise remain invisible (Zeng 2015). Probabilistic simulation signifies a shift from representative democracy toward a more predictive model of governance, where representation is verified not by the expressed political will of the populace, but by the flow of latent long-term and planetary-scale patterns.

Autonomy of Algorithmic Democracy
Calvin, as a central interpreter of this predictive governance model, conclusively revisits the philosophical tenets firstly brought up in “Evidence”. She suggests that humanity has never been in command of its own destiny, stating that “it was always at the mercy of economic and sociological forces it did not understand—at the whims of climate, and the fortunes of war” (Asimov 1950: 218). For Calvin, the Machines cannot take away human agency, as free will never fully existed; instead, they formalize constraints that were already present, and can only dispel the illusion of freedom as a sort of historical purpose.

Yet, such determinism does not need to be categorically opposed to agency or free will. For example, philosopher of mind Daniel Dennett’s compatibilism (2003) posits that free will remains real, though it is not based on any natural law or divine power. Rather, it is an evolved construct born of human activity and beliefs (p. 13). Dennett’s argument explains why people might advocate for free will as though it “were a political condition that might be under threat, might spread, or go extinct as a result of what people came to believe” (ibid.). For Dennett, Calvin’s determinism does not suffice to eliminate agency, as free will operates within constraints, developing as an evolutionary response to changing cultural environments.

However, both Calvin’s hard AI determinism and compatibilist views like Dennett’s does complicate democratic ideals of free choice, particularly in elections, where voters assume they are freely selecting among alternatives (McKenna 2024). In Asimov’s story, where the Machines are themselves capable of determining who their Co-Ordinators should be, it hardly makes sense to acknowledge elections like Byerley’s as truly democratic. However, there is one exception: an ideal algorithmic democracy could still be democratic if the evolutionary direction of the capacities for free will has now been transferred to the Machines, which constitutes a possible scenario if one follows evolutionary compatibilism á la Dennett.

Calvin’s final words to Byerley capture the core of the determinist dilemma: “Think, that for all time, all conflicts are finally evitable. Only the Machines, from now on, are inevitable!” (Asimov 1950: 218). On the one hand, Asimov’s Machines resolve the constraints that have curtailed human agency, but they do so by becoming an unavoidable and absolute principle that sidelines the centrality of humanity’s own willings. On the other hand, as the Machines are endowed with an autonomy derived from their reinterpretation of the Laws, they do have the ability to not act as autocrats, but as facilitators of collective human will.

The story’s Regional Co-Ordinators demonstrate a form of compatibilist agency, managing the Machines’ deliberate errors and adapting decisions to balance regional needs and cultural plurality. Each region presents singular challenges: in the East, yeast production is modulated according to fleeting trends in flavor, while the Tropics face labor shortages despite the Machines allocating more workers to match rapid economic and population growth. Europe, resembling a post-imperial Greece, stagnates in its response to global shifts, whereas the North, still viewing itself as the world’s center, seeks to uncover untapped data to extend the Machines’ range. Through these regional distinctions, the Regional Co-Ordinators localized agency is compatible with the deterministic framework of the Machines. Though voting and electoral processes are diminished, democratic participation is not obsolete—humans are still called to adapt to new constraints, opening pathways for societal evolution and autonomy.

Listening to Planetary AI
Now, the rise of planetary-scale computation calls for a reconsideration of both the mechanics of governance and the future of democracy itself. As philosopher Benjamin Bratton (2022) points out, humanity has not built a single, giant computer, akin to Asimov’s Machines. Rather, they came to produce a “massively distributed accidental megastructure” (ibid.), as an interconnected, computational sphere which spans continents and oceans. This megastructure rationalizes the information flows of global governance in ways never intentionally designed.

While the political implications of Planetary AI are more unpredictable than Asimov’s vision, his story does present a rather complex geopolitical trajectory. Without eliminating existing democratic institutions and representatives, it redefines politics on a more intricate level. In this future, politics is no longer primarily about responding to the populace’s articulated interests or opinions. Instead, it revolves around ‘listening’ to diverse identities, preferences, and aspirations. Democracy shifts from being a system of voting mechanisms and divisions of power to an almost organic condition in which humanity—embedded within massive data flows—becomes integrated with the very fabric of governance, albeit passively. This model of governance calls for a practice of ‘machine listening’ that synthesizes collective human interests and opinions with the planetary demands of ecological and economic systems. This participation in democracy—if still to be called democracy—is one of adaptation, where the alignments between human and machine are recalibrated to complex and evolving realities.

The Machines’ post-representative strategy—attending to larger patterns of life rather than individual demands—prefigures contemporary AI governance experiments. This organic, albeit also quasi-totalitarian, conception of democracy especially finds its real-world analogs in Japanese localized experiments. The activist and wrestler Michihito Matsuda, for instance, has promoted an AI-led approach to direct democracy since 2018 with the virtual “AI Mayor” candidate in Tama City. In his manifesto, he desires for The AI Mayor to amplify individual voices in real-time without traditional voting, concretely revoking the legend Prince Shōtoku (574-622 CE), who supposedly could hear ten petitioners simultaneously (Matsuda 2018).

More concretely, Takahiro Anno’s campaign for Tokyo governor in 2024 also showcased a listening approach. A trained AI engineer and science fiction author, Anno employed a strategy based on “broad listening” (in opposition to traditional “broadcasting”) and used an AI avatar of himself to interact with voters, creating a ‘weak’ form of virtual politician. Anno’s AI-assisted campaign, without resorting to full AI representation, garnered 154,000 votes—setting an all-time high for a new candidate without a political background (Plurality Tokyo 2024). The Plurality community, where Anno’s concept is rooted, argues that while modern broadcasting methods have evolved in mass democracies through radio, television, and digital platforms, broad listening—synthesizing multiple perspectives—remains underdeveloped and overlooked (Tang et al. 2024). This opens possibilities for democratic representation, where political decisions can involve interpreting and acting upon transversal patterns rather than the formalizing of opinion groups as mandates of a parliamentary puzzle.

The ‘listening’ dilemma here complexifies Asimov’s universalism critique on how Planetary AI can serve humanity’s safety. Indeed, one can question how such a system will listen to the complexities of diversity without amplifying mere tendencies? Can a model optimizing the long-term planetary order also accommodate the subtle, often contradictory forces defining life, or will it flatten humanity’s pluralities into variables of an ultimately diffuse equation?

The compact spectacle of algorithmic representation: “Franchise” (1955)

Watching the 1952 U.S. presidential election on television for the first time, Asimov found himself disappointed by a scenario that seemed like something out of one of his own stories. While Asimov had voted for the Democratic candidate Adlai Stevenson, a computer named Univac called the victory for Dwight Eisenhower after analyzing a mere 5.8% of the count. What troubled Asimov was not the actual outcome, but his experience that the computer had stripped away all suspense and human engagement from the election event. Univac, Asimov reflected in his diary, had rendered the show on TV “no fun anyway” (Asimov 1979: 663). Even though Asimov would have been sure to have had his idiosyncrasies stifled by hearing dramatic references in the CBS broadcast to an “electronic brain” and “mechanical monster” (Chinoy 2024), the plotting of hope and anticipation for representation was called off early.

While prediction and statistics in electoral forecasting were nothing new, the actual use of computers for this role marks a historical turning point, also giving unprecedented public exposure to computers (ibid.). The spectacle surrounding ‘the Univac moment’ exemplifies what Bernard Stiegler has called “telecracy”—the degradation of public opinion into a passive audience, where marketing supersedes elected delegates in representing the social whole (Stiegler 2010: 3). Asimov, realizing the larger ramifications of this convergence between computers and democracy, found himself entertaining a possibly more suspenseful scenario: What if computers didn’t merely predict elections, but actively determined their outcomes?

Super-Representation and Hyper-Citizenry
In “Franchise,” Asimov explores this plot by presenting a future where the United States has, by 2008, rebranded itself as an “electronic democracy.” Instead of millions casting votes, the supercomputer Multivac selects one statistically representative citizen to vote on behalf of the entire electorate. This (always male) individual ‘votes’ by responding to a series of questions from Multivac, which then finetunes its vast societal data to calculate both an election result and policy directions for the nation. From the human perspective within Franchise’s narrative, what is herein recognizable of democracy is only its “guessing game”. It is fully impossible to know how or why Multivac selects any given individual as the voter of the year, ensuring that algorithmic democracy leaves space for the stream of folk rituals surrounding elections.

Thus, algorithmic democracy marks a transition from biopolitics’ post-individual population management into the strange contradiction of an “apparent individualization of statistics” (Stiegler 2016: 109), which personalization form can be personified by the story’s protagonist Norman Muller. Muller, a seemingly average citizen from Bloomington, Indiana, is the one chosen as ‘Voter of the Year’. His name is a deliberate choice, reflecting a statistical banality like the ‘normal citizen’ of a vox pop. This parallels the naming logic behind Leader Lars of The Synthetic Party, who echoes a range of societal biases where, for instance, more Danish CEOs are named ‘Lars’ than there are female CEOs at all (Herrie & Staunæs 2024). When the individual represents the collective will, ‘he’ abstracts the biases inherent to any system of majority rule.

The demands of majority representation is evident in Muller’s psychological turmoil, as he grapples with the responsibility of embodying the entire electorate—of being a hyper-citizen. The security and secrecy around being the chosen voter is overwhelming, and Muller himself becomes a public figure. Muller’s individual anxiety—“They can’t make me vote if I don’t want to. I’ll say I’m sick. I’ll say—” (Asimov 1955: 10)—underscores the electoral disillusionment experienced by citizens of mass democracy, where the continuity of participation is reduced to simulative probing mechanisms such as polling.

Thus, Asimov’s story critiques not just the right to vote, but also the rite of voting itself—the way that elections are idealized as conservative rituals focused more on affirming the status quo than facilitating political agency. Norman Muller is effectively drafted into this electoral service, which resonates with Asimov’s earlier experiences of duty, as argued in my previous analysis of “Evidence”. Like military conscription, voting is not a choice but an obligatory act of civic service—it is a ‘franchise’, implying a privilege granted by the state.

A concurrent analogy can be found in Bertolt Brecht’s famous poem Die Lösung, written in response to the 1953 East German uprising but only published posthumously in 1959. In the end, Brecht sardonically suggested that it would be more efficient for the government to dissolve the people and elect a more compliant populace (Brecht 1959; 2003). While Asimov and Brecht were not directly influenced by each other, their satire aligns in proposing hyperbolic solutions where governmental control reduces or negates the electorate’s agency. This coincidence highlights a broader erosion of participation in liberal post-war governance.

Immanent critique beyond dystopia and utopia
“Franchise” has been described as “the most dystopian view of democratic representation” by Jan-Werner Müller, a renowned theorist of democracy and populism. (Müller 2021: 83). In Democracy Rules, Müller contends that “Franchise” illustrates the reduction of elections to its element of census, which forgets that representations should be thought of “not as neutral descriptions of some obvious, let alone objective, social reality, but as calls to battle” (ibid.) with “creative and dynamic aspects” (p. 85). However, the dystopian label does not account fully for the intricacies of Asimov’s creative battle against representation, as the story readily acknowledges how Multivac has made mistakes in its assessment of social reality (Asimov 1955: 10). Rather than positing any objective theory of representation, Asimov’s critique especially singles out how it is really the anti-political undercurrents of democracy—refusal, disengagement, and even irreverence—that is transferred away from formal democracy by Multivac’s formalization of the electoral vote.

Norman Muller’s father-in-law, Matthew, reflecting the discontent of the older generation, laments that Multivac has taken away both the right not to vote and the pleasure of voting irresponsibly: “I can’t vote cockeyed just for the fun of it anymore!” (Asimov 1955: 7). Matthew highlights what superintelligent Multivac has probably reasoned to be the central inefficiency of democracy: that not everyone votes, and many who do vote are disengaged. Under Multivac’s rule, it is actually the problem of mass voter apathy that has been ‘solved’ by concentrating the voting into a super-representative individual to bear the burden for all.

Within any political democracy, there is always a dissolved human populace lurking existing structures—those who, as Brecht’s poem suggests, are due to be dissolved for incompliance. Multivac’s dialectical negation of exclusion and disenfranchisement mirrors the objective of The Synthetic Party, which claims to represent the 15-20% of voters who always abstain from Danish elections—a figure that doesn’t even include the nearly 10% of adult residents who are excluded from voting altogether, because they are not ‘Danish’ (Nørtoft 2019), as well as the blank or deliberately ineligible votes.

Akin to how Multivac operates with vast societal datasets to identify its uttermost representative voter, The Synthetic Party uses large language models to synthesize political programs from the 200+ registered micro-parties that remain off the parliamentary ballots. The party leverages this fine-tuned dataset to create a voice out of those otherwise left outside the formal democratic process (Panarari 2022; Xiang 2022).

Entertaining the idea that AI models like Multivac and The Synthetic Party might grasp a fundamental feature of democracy, tech blogger Pablo Haya has recognized the etymological brilliance of algorithmic representation, as it “aims to convert that 20% of idiṓtēs who do not vote—understanding the term as the ancient Greeks did, for people uninterested in public affairs—into 20% of modern idiots who end up giving their vote to an AI” (Haya 2022, my translation). In this light, political AI raises the question of whether democracy’s reliance on citizen representation is, in some ways, already absurd. As Asimov’s Franchise suggests, democracy’s performative elements may be stretched to the point where the system itself becomes something entirely different—something that might still holds space for creativity and unpredictability, but only in new and unfamiliar forms, as seen in Multivac’s super-representation and The Synthetic Party’s amalgam of the disenfranchised.

Algorithmic spectacle as the collapse of representation
Asimov’s vision of algorithmic democracy resists simple classification as dystopian or utopian. Rather, it speculates on how democracy might be reconfigured at the intersection of spectacle and technology. Borrowing from Guy Debord’s classifications of the “society of the spectacle,” “Franchise” marks the onset of a more integrated and compressed form of social performance and image politics. Debord traced modern society’s evolution through different phases of spectacle—from the “concentrated” form found in authoritarian regimes to the “diffuse” version of post-war consumerism, ultimately culminating in the “integrated” form characteristic of neoliberal biopolitics (Debord, 1988; 2009: 8-9). “Franchise” specifically reflects the intersection between a diffuse spectacle of mass democratic marketing with the concentrated authoritarianism of bureaucracy that consolidates a rule of information control. This inaugurates an integrated model in which state and corporate interests blend seamlessly, even in the sacred act of voting.

Norman Muller, as a statistical stand-in for the entire electorate, personifies this integration; the Western individual, finally alone, is in a service where ritualized theatricality conceals the underlying mechanisms of capital and information. The sociology theorist Jean Baudrillard’s theory of mass opinion has here concretized Debord’s spectacle: According to Baudrillard, the masses absorb information like entertainment, but provide no meaningful feedback in return. The political simulation can persist because the mass electorate acts like a ‘black hole’—absorbing input, yet offering no response, accumulating density without transmission (Baudrillard 1978: 89). In this view, Muller’s super-representation reflects how democracy approximates a dying star, wherein the ever-heavier system eventually draws everything in.

In some way, political AI harnesses Baudrillard’s thesis on the impossibility of representing the silent majorities, as the algorithmic representations of Multivac and The Synthetic Party fundamentally diverge from the logic of liberal democratic representation. Rather than trying to convert non-voters into active participants, these systems represent non-participation itself, creating a voice of the silent majorities rather than for them. Baudrillard’s critique of polling captures this evidently: “polls can never represent anything at all, as a word can be of a thing, an image of a reality, or a face of inner feelings (…) it is a total misunderstanding to apply the logic of representation to a system built on simulation” (Baudrillard 1983: 115-116). While the historical electoral system could claim to be representative by staging a dialectic between representatives and the represented, Baudrillard argues that this can no longer hold true in the age of simulation: “The model, unlike the concept, is not of the order of representation, but of simulation,” (ibid.) and thus the dialectic collapses into dissuasive and nonreferential surface.

In Baudrillard’s sense, Multivac and The Synthetic Party cannot be generating any form of “real” representation—what they do is materialize absence. Like ‘the government’ of Brecht, simulative politics entails a task of creativity through absence, as a forming of the very void where agency is dissolved and reconstituted. By integrating dynamics of mediated inclusion and exclusion into its model, political AI negates any facsimiles of vox populi, vox dei. As a variant of the negative machine agency (Parisi 2022), the crux is no longer about representing presence but carving out space for non-participation, refusal, and creation through absence.

The collapse of societal spectacle is thus further compressed with anti-political, AI-led parties like The Synthetic Party. While “Franchise” of 1955 marked an early transition toward integrative spectacle where representation remained separated into technical efficiency and media-driven involvement, The Synthetic Party is on the move toward an even more compact stage—from neoliberal biopolitics to algorithmic democracy. Here, stochastic language models embody the chaos and contradictions that Multivac’s formal optimization left aside for humanity’s ritualistic engagement in the media. Rather than sidelining these rites, The Synthetic Party practically embeds the performative and symbolic layers of engagement within its AI models.

Immanent Critique as Methodology for Political AI

The analyses have positioned Asimov’s stories on political AI as works of social criticism, and drawn on theoretical writings to probe the possibilities for critique in his work. Asimov, who in his early youth took part in the Marxist science-fiction fan collective of “Futurians” (1937-1945), can be interpreted to develop an ‘immanent’ form of criticism which frames technological dilemmas to reveal societal contradictions3. However, Fredric Jameson have argued that Asimov’s conception of the ‘societal’ remained at the inception of a nascent sociological period within science-fiction, as a modulation of ‘social satire’ which Jameson deemed too parochial and limited for truly qualifying as critique (Jameson 2005: 91-92).

While Jameson could dismiss the breadth of Asimov’s immanent critique, the real-world “materialization” of his stories in political campaigns underscores how a “defamiliarized present” can remain confined within its parochial limits, to borrow Jameson’s analytical terminology. As Jameson argues, science fiction like Asimov’s “does not give us ‘images’ of the future… but rather defamiliarizes and restructures our experience of our own present” (Jameson 2005: 286, his italics). Today, the emergence of political AI—within a future that was, for Asimov, still highly speculative, but which unfolds from his time—demonstrates the persistence of Asimov’s social satire, or (quasi)sociological approach. This new present, thus, takes part in a perpetual farce. Paraphrasing Marx, one can entertain that it was maybe not the specter, but indeed ‘the spook’ who has been haunting the political economy (Staunæs 2023).

For political AI to be conceived as a form of immanent social critique, I have pursued a core distinction: that the political AI stories and experiments can be actively skeptical of societal development as a whole (in this case, algorithmic democracy) while maintaining an optimism towards the critical potential for their specific choice of counter-maneuver (such as the virtual politician) to possess the spectral capacity of consisting both with and against the system. To redefine concepts of García-Marzá and Calvo (2024), this frame of distinction casts ‘virtual politicians’ as revealing the inherent contradictions of supercilious ‘algorithmic democracy’.

The methodology of immanent critique contributes a probing into the intellectual origins of algorithmic democracy, and also fosters an imaginative framework for assessing the interplay between AI and politically democratic systems. Increasingly, Asimov’s science fiction can be understood not merely as predictive exercises, but as speculative examinations of algorithmic power formation within democratic contexts, where the persistence of satirical and parodical tendencies reflect the political world’s long-standing entrapment within parochial limitations.

Closing Reflections

As the creator of The Synthetic Party, I have been caught up in the peculiar play of navigating an Asimov-like world. The party’s inception in the Danish tech-hub “MindFuture” (Life with Artificials, 2022) placed me at the helm of a project that spans media, academia, and the art world. Yet, the further I ventured into this reality, the less I felt like its architect. Instead, I have been more like Stephen Byerley, steering a spectacle of identity that straddles the line between the familiar—the ‘meat robots’ of modern human condition—and the alien logic of ‘stochastic parrots,’ embodied in AI models.

In many ways, where Asimov incubated the imaginary of algorithmic democracy, we at The Synthetic Party have ignited its anti-political inclination. Unlike political systems that negate through exclusion—casting non-voters outside electoral processes—The Synthetic Party has built upon Multivac’s dialectical negation by embedding disenfranchised and abstentionists. Political AI, as realized through today’s AI-driven parties and virtual politicians, constitutes the most honest although extreme expression of technology-infused political agents. It pushes democracy toward its own logical extremes, revealing the absurdities of representation by making it both more direct and more surreal. In its ultimate teleology, political AI does not care about replacing human representatives; it strives for a ‘listening’ that, while alien, might be more in sync with the infrastructural logic of data-driven, planetary-scale decision making.

The main theoretical contributions of analysis has been the following:

1) Laws-and-Algorithm:
Asimov’s “Evidence” prefigures current debates on virtual politicians by questioning whether algorithmically encoded ethics can resolve democratic failures. The Three Laws of Robotics, while reflecting universalist promises of ethical governance, also inherit fundamental labor abstractions of algorithmic systems. This duality—between the universal abstraction of laws and the social abstraction of algorithms—raises questions on the programmability of politics.

2) Democratic Compatibilism and Machine Listening:
“The Evitable Conflict” raises to a planetary scale, where AI systems, through probabilistic modeling, facilitate governance by absorbing human will into their predictive frameworks. In Asimov’s vision, democracy is not simply diminished but potentially transferred to machines, who interpret society’s program to facilitate collective will. This ‘machine listening’ renders politics to be evaluated through long-term patterns of psychological and planetary behavior.

3) Rites of Algorithmic Representation:
In “Franchise”, Asimov reduces the right of voting itself to computational processes, laying bare the persistence of ritual in democracy. This spectacle of representation is shown as more compressed by AI-driven parties, where representational creativity merges with the cultural archive of large language models. Here, the non-participation of AI is amplified, as the whole theatricality of politics enables a simulation that creates through absence rather than presence.

Conclusion
Asimov incubated a vision of political AI that reveals itself as an exposition of democracy. As political representation collapses into algorithmic spectacle, the lines between control and liberation, automation and autonomy, becomes intangible. Here, Asimov‘s stories prompts one to consider: Are the distortions of political AI truer to the reality of representation than the polished facades of publicity that typically are passed off as ‘democratic’? If so, the bitter lesson (Sutton 2019) for society might be that the algorithm of democracy can firstly be run with sufficient compute.

/ Asker Bryld Staunæs, on behalf of Computer Lars and The Synthetic Party

Literature

  1. The character Stephen Byerley, later confirmed by Asimov to be envisioned as a robot, also marks the earliest incarnation of R. Daneel Olivaw (Asimov 1991, p. 17) who is often credited as Asimov’s first humanoid robot. From Byerley to Olivaw, Asimov develops the humanoid robot as a central political agent in a cosmic scheme that ranges from Olivaw’s covert role as the First Minister Eto Demerzel in Foundation Trilogy (1951-53) to really foundational leadership in the late integration of Asimov’s Foundation and Robot-series from 1986 where Olivaw shapes the galaxy towards a formation of collective consciousness. The legacy of Olivaw even reaches the absolute stars—serving direct inspiration to Mr. Spock, Star Trek’s all-time political sci-fi icon (ibid., p. 20). ↩

  2. Philosophically, such metaphors harks back to at least Thomas Hobbes’ Leviathan (1651), where the commonwealth is referred to as an “Artificial Man”, and to treaties like The Oldest Systematic Program of German Idealism (1795) where the callow triumvirate of Hegel, Hölderlin, and Schelling described the State as a “machine” treating free humans as “cogs”. ↩

  3. The immanent rather than transcendental characteristics of Asimov’s criticism is likely caused by his framing of robots as machines rather than metaphors. In the introduction to Robot Visions, Asimov distances his robot stories from earlier fearful conceptions, such as Karel Capek’s R.U.R. (1920), where robots symbolized slave rebellion, and Mary Shelley’s Frankenstein (1818), where the man-made monster posed a metaphor for the dangers of scientific transgression. From Asimov’s perspective, these represent a “Frankenstein complex” (Asimov 1991, p. 13) where the robot et al. symbolizes transcendental class interests, or a political agenda. ↩