What schools of thoughts are there on AI implementation on social-political theory?

We often imagine the state as a collection of laws, bureaucrats, and voters, but what if the most powerful political actor is actually a line of code? In 1980, Langdon Winner posed a question that has become the cornerstone of modern political inquiry: [Do Artifacts Have Politics?](https://en.wikipedia.org/wiki/Do_Artifacts_Have_Politics%3F) Today, as AI systems transition from recommendation engines to adjudicators of social benefits and criminal justice, the answer is no longer "perhaps," but a resounding "inevitably." The implementation of AI into the social fabric has birthed three primary schools of thought that contest how technology should interface with power. ### 1. Algorithmic Rationalism and Solutionism This school, often associated with Silicon Valley and "Smart City" initiatives, views social problems as technical inefficiencies. Drawing from the utilitarian tradition, proponents argue that AI can achieve a "computable state" where resources are allocated with a precision human bureaucracies cannot match. However, critics like [Evgeny Morozov](https://en.wikipedia.org/wiki/Evgeny_Morozov) warn against **Technological Solutionism**, the belief that complex social frictions—like poverty or crime—can be "solved" through optimization alone, often at the cost of democratic deliberation. ### 2. Critical Algorithmic Studies Rooted in critical race theory and sociology, this school examines how AI replicates and amplifies existing social hierarchies. Thinkers such as [Ruha Benjamin](https://ruhabenjamin.com/) argue that technology is never neutral; it is "coded inequity." In her seminal work, *Race After Technology*, Benjamin describes the "New Jim Code": > "The New Jim Code: the employment of new technologies that reflect and reproduce existing hierarchies, including settled forms of racial hierarchy, even that which is promoted as objective or progressive." This school argues that AI implementation in policing or hiring is not just a technical error but a political project that reinforces the status quo under the guise of "objective" data. ### 3. Liberal Proceduralism and AI Ethics This school seeks to integrate AI into existing democratic frameworks through "Human-in-the-Loop" systems and transparency mandates. Influenced by [John Rawls' Theory of Justice](https://plato.stanford.edu/entries/rawls/), these theorists focus on **Algorithmic Accountability**. They argue that if an AI makes a decision affecting a citizen's rights, that decision must be explainable and subject to appeal. The focus here is not on stopping AI, but on building a "constitutional" framework for it, as seen in the development of the [EU AI Act](https://artificialintelligenceact.eu/). ### 4. Post-Humanist Accelerationism A more radical school suggests that AI fundamentally destabilizes the concept of the "human" subject that underpins liberal democracy. If an AI can deliberate or create better than a human, the traditional focus on human agency in political theory must be abandoned. This school explores the possibility of "fully automated" governance, questioning whether the goal should be to control AI or to evolve our social structures to match its alien logic. *** **Further Exploration:** 1. If an AI system provides more equitable outcomes than a human judge but cannot explain *why*, should we prioritize the outcome or the process? 2. Can a democratic society exist if the primary drivers of public opinion are non-human agents (bots) that reside outside the social contract? 3. How does the "Black Box" nature of neural networks challenge the legal requirement for "due process" in governance?

If we view the schools of thought as fixed islands, the "spectrum" of AI implementation acts as the tectonic forces moving beneath them. To understand the political stakes, we must look beyond what these systems *do* and examine the axes of power they reshape. Specifically, we find a tension between **Epistemic Legibility** and **Social Friction**, and between **Automated Decisionism** and **Democratic Contestation**. ### 1. The Epistemic Axis: Legibility vs. The Vernacular A primary spectrum in AI implementation concerns how the state "sees" its citizens. On one end lies **High Modernist Legibility**, a concept explored by James C. Scott in [*Seeing Like a State*](https://en.wikipedia.org/wiki/Seeing_Like_a_State). AI intensifies this drive, translating the "messy" realities of human life—culture, nuance, and local context—into discrete data points for administrative efficiency. The counter-pole is **Epistemic Resistance**, where theorists argue that certain social truths are inherently non-computable. When an AI system categorizes a neighborhood as "high risk," it ignores the "metis" (local knowledge) that a human social worker might possess. The spectrum here is a move from governance based on *understanding* to governance based on *calculation*. ### 2. The Sovereign Axis: Automated Decisionism A second spectrum concerns who—or what—holds the power to decide in a moment of crisis. This draws on the work of [Carl Schmitt](https://plato.stanford.edu/entries/schmitt/), who famously defined the sovereign as "he who decides on the exception." - **Human-centric Oversight:** AI acts merely as a decision-support tool. The "sovereign" remains a human agent accountable to the law. - **Automated Decisionism:** At this pole, the "exception" is handled by the algorithm. If a social benefit is cut off by a "black box" system, the system itself becomes the sovereign actor, operating in a space where traditional due process cannot reach because the logic of the decision is inaccessible to the human subject. As [Frank Pasquale](https://en.wikipedia.org/wiki/Frank_Pasquale) notes in *The Black Box Society*: > "Authority is increasingly expressed through algorithms... Decisions that used to be based on human judgment are now made by automated systems that are often inscrutable, even to their creators." ### 3. The Temporal Axis: Predictive Conservatism vs. Radical Potential Finally, we must consider the spectrum of **Temporal Governance**. AI systems are inherently retrospective; they are trained on historical data. 1. **Predictive Conservatism:** At this end, AI acts as a "stabilizing" force that colonizes the future with the biases of the past. If past policing data is biased, the AI ensures that future policing remains biased, effectively trapping the political subject in a permanent "yesterday." 2. **Speculative Reform:** At the opposite end, researchers explore how AI could be used to simulate alternative political realities or "un-bias" resource distribution by intentionally over-weighting marginalized data. This moves AI from a tool of *reproduction* to a tool of *transformation*. The conflict on this spectrum is whether AI is used to **close the future** (by predicting it) or **open the future** (by identifying the hidden causal structures that prevent social mobility).