AI governance is disrupting the foundations of society

The study warns: As AI expands its role in mediating human behavior, the space for political action, as envisioned by thinkers like Hannah Arendt, may shrink beyond recognition. When human action is framed only as data to be predicted, it ceases to be action at all; it becomes behavior to be managed. In such a world, the boundary between the personal and political dissolves, and the public sphere collapses into a stream of behavioral data.

The rapid integration of artificial intelligence (AI) into every aspect of society is provoking profound transformations in how human collectives are governed. A new editorial titled "Age of Disruption," published in AI & Society, lays bare the philosophical and political stakes of this transition.

Authored by Jan Soeffner, the piece argues that as AI-driven data infrastructures increasingly displace traditional institutions, we may be witnessing the end of politics as the primary mode of societal organization.

What forces are driving the shift from politics to technicity?

The editorial identifies a growing trend in which traditional forms of governance, public bureaucracies, democratic institutions, and representational politics, are being superseded by technical infrastructures. These systems, built on data extraction, predictive algorithms, and behavioral governance, no longer require the pluralistic negotiation characteristic of political systems. Instead, they operate through real-time optimization and feedback, replacing deliberation with automation.

Soeffner links this shift to historical tensions between communal and planetary scales of governance. Drawing on Marshall McLuhan's concept of the "global village," the editorial contends that while digital technologies enable networked communities, they also erode the societal frameworks that allow pluralistic coexistence. AI, empowered by predictive analytics and ubiquitous connectivity, now administers society not through public discourse but through algorithmic regulation. This marks a pivotal moment where politics risks being reduced to a legacy function, overshadowed by technicity.

The central problem, according to the editorial, is that engineering logics, especially those underpinning AI, are not designed for the messiness of human pluralism. When technical systems treat social issues as problems to be solved rather than conflicts to be negotiated, they risk suppressing dissent, flattening values, and eliminating the unpredictability that gives democratic societies their resilience.

What are the dominant narratives shaping AI governance?

Soeffner critiques three prevailing visions of AI-enabled social organization. The first is the techno-utopian ideal from Silicon Valley, rooted in countercultural ideals of decentralized empowerment and digital commons. While promising spontaneous self-organization, this narrative often masks corporate surveillance and capitalist control. The second is the model of the benevolent smart state, which uses AI to nudge, gamify, and sanction citizens for collective benefit. Though efficient, it tends toward technocratic paternalism and risks drifting into totalitarianism.

The third narrative is the network state concept, advanced by crypto-libertarians, which envisions replacing nation-states with crowd-funded micro-communities based on ideological homogeneity and digital sovereignty. Although appealing to some as a model of autonomy, this framework reproduces exclusionary logics and undermines the social contract.

Each of these narratives, the editorial argues, mischaracterizes society as a solvable problem rather than a dynamic, contested process. They fail to account for cultural pluralism, historical struggle, and the role of conflict in democratic innovation. As AI becomes more embedded in societal governance, these narratives provide inadequate blueprints for equitable and accountable futures.

Can AI governance coexist with pluralistic society?

The study warns: As AI expands its role in mediating human behavior, the space for political action, as envisioned by thinkers like Hannah Arendt, may shrink beyond recognition. When human action is framed only as data to be predicted, it ceases to be action at all; it becomes behavior to be managed. In such a world, the boundary between the personal and political dissolves, and the public sphere collapses into a stream of behavioral data.

Moreover, AI's reliance on supposedly value-neutral optimization often conceals embedded biases, which disproportionately affect marginalized groups. Attempts to regulate these biases have so far been technocratic and incomplete. More dangerously, algorithmic governance may empower both autocratic regimes and micro-totalitarian enclaves by granting them unprecedented tools for control.

Despite these challenges, the editorial insists that the future is not predetermined. Acknowledging the limitations of technical solutionism is a first step toward reclaiming human agency. The task ahead is not merely to regulate AI, but to reintegrate political discourse into the design of AI systems. This includes preserving pluralism, enabling dissent, and ensuring that technological power remains subject to democratic deliberation.

If society is to truly endure this age of disruption without sacrificing its foundational values, it must find ways to make AI answer to politics, not the other way around, the author concludes.