TLDR: The increasing integration of Artificial Intelligence into governmental roles is sparking global discussions on governance and accountability. Albania and Japan are at the forefront, with Albania utilizing a digital assistant, Diella, for procurement and Japan’s Path to Rebirth party proposing an AI leader. These developments highlight a shift from AI as a hidden ‘decision support’ tool to an explicit institutional actor, raising concerns among economists about potential ‘techno-feudalism’ and the opacity of AI’s statistical inference-based decision-making.
The rise of Artificial Intelligence (AI) is ushering in a new era of governance, prompting nations to confront the profound implications of integrating advanced algorithms into public administration. This shift is moving beyond the traditional role of AI as a ‘decision support‘ tool, with some countries now assigning explicit institutional roles to AI systems. This development has sparked discussions among policymakers and economists, with concerns ranging from potential AI ‘domination‘ to the emergence of ‘techno-feudalism‘.
Recent developments in Albania and Japan exemplify this trend. On September 19, 2025, Albania’s AI minister delivered her inaugural speech in parliament, marking a significant step in formalizing AI’s role in government. The Albanian government has officially tasked its digital assistant, ‘Diella,’ with managing procurement processes. Similarly, Japan’s small ‘Path to Rebirth‘ party has announced its intention to appoint an AI as its leader. While these instances do not represent a complete transfer of authority to machines—Diella remains a supervised workflow tool, and the Japanese party still requires a human representative for official filings—they are seen as pivotal. These moves transform algorithmic decision-making from a background function into a recognized, public institutional role, signaling that algorithmic governance is becoming explicit and necessitating a re-evaluation of institutional design, legitimacy, and accountability.
Historically, algorithms have long been involved in governance, determining aspects such as job advertisement reach, tax audit flags, welfare case prioritization, and police patrol scheduling. However, these functions were largely hidden under the guise of ‘decision support.’ The current wave of AI systems, characterized by their ability to learn from data, adapt over time, and operate at scale, introduces a new dynamic. Unlike earlier rule-based systems that executed fixed instructions, modern AI generates patterns, ranks alternatives, and can propose unforeseen actions, making them powerful yet harder to scrutinize.
Ozan Ahmet Cetin, writing for TRT World on September 30, 2025, highlights that this shift means governing with systems that evolve, where the reasoning behind their decisions must remain intelligible to preserve democratic oversight. The article draws parallels to Enlightenment thinkers like Leibniz and Condorcet, who envisioned replacing disputes with calculation, and Jeremy Bentham’s utilitarian policy aimed at maximizing collective happiness through rational computation. Contemporary algorithmic governance appears to actualize this vision, promising decisions free from whim and prejudice.
However, this evolution also brings challenges. Max Weber’s concept of the ‘iron cage‘ of bureaucracy, characterized by formal procedures and hierarchical oversight, becomes even tighter with algorithmic systems. While they promise consistency by removing discretion and enforcing uniformity, they also introduce opacity. Modern AI operates on statistical inference rather than explicit logic, mapping complex correlations in data. This allows for flexibility and adaptation but makes it difficult for policymakers to explain specific recommendations or reconstruct the chain of reasoning, effectively replacing ‘visible bars with invisible ones.’
Furthermore, AI enables micro-differentiation, allowing risk scores, eligibility decisions, and policy nudges to be tailored to neighborhoods or individuals, a level of granularity impossible with earlier administrative systems. While this offers opportunities for precise resource targeting and reduced inequity, it also risks fragmenting the idea of a public by replacing collective treatment with individualized optimization, complicating political justification.
The continuous, real-time operation of modern AI systems, which ingest data and adjust decisions on the fly, contrasts sharply with the periodic and retrospective nature of classic bureaucratic systems. This ‘dynamic governance‘ complicates oversight, as the constant updates make it challenging to evaluate what exactly is being audited or reviewed.
AI-led governance also redefines agency, as these systems can generate options and propose strategies unforeseen by their designers, blurring the line between decision support and decision-making. Public officials must now govern not only populations but also the models themselves, learning when to trust, override, and translate public values into technical parameters. The critical question remains whether the AI model keeps learning in a way consistent with democratic intent.
Also Read:
- New AI Personhood Framework Proposed for Governing Agentic Artificial Intelligence
- Beyond the Hype: Unpacking the Overlooked Aspects of AI Accountability and Governance
The experiments in Albania and Japan, whether intentional or not, have made algorithmic governance visible. They serve as crucial early case studies for designing the necessary norms, audit practices, and legal frameworks to ensure algorithmic decision-making remains legitimate, contestable, and aligned with democratic principles before it becomes deeply entrenched.
