Algorithmic Decision-Making In Indian Administration: From Formal Authority To Substantive Control And Accountability
Awantika Neeraj, National Law University of Meghalaya
ABSTRACT
Artificial intelligence and algorithmic systems have migrated from the periphery of Indian public administration into its operational core, determining who receives a ration card, how a tax return is assessed, and whom a policing algorithm flags for surveillance. This paper argues that this shift has produced a structural mismatch between Indian administrative law doctrine, which is built on the premise of an accountable human decision- maker exercising formal authority, and the operational reality of substantive control now exercised by opaque, often unaccountable algorithmic systems. Using Aadhaar-linked welfare delivery, the Faceless Assessment Scheme under the Income Tax Act, and facial-recognition-based protest surveillance in Delhi as case studies, the paper demonstrates how algorithmic mediation strains foundational doctrines the duty to give reasons, the principles of natural justice, and the rule against fettering of discretion and produces outcomes that are difficult to challenge precisely because their underlying logic is neither disclosed nor easily traceable to an identifiable human decision-maker. The paper evaluates whether India's principal data- protection statute, the Digital Personal Data Protection Act, 2023 and the Digital Personal Data Protection Rules, 2025, meaningfully close this gap, and finds that it does not: Indian law confers no right against solely automated decisions and no right to an explanation, unlike the European Union's General Data Protection Regulation. Drawing on the Supreme Court's 2026 Draft Regulations for Use of Artificial Intelligence in Courts as an emerging, if narrow, domestic model of "human primacy," and on Canada's federal Directive on Automated Decision-Making as a comparative template, the paper proposes a framework of substantive algorithmic accountability for Indian public administration built on four pillars: mandatory algorithmic impact assessments, a statutory duty of explainability, guaranteed human review in high-stakes decisions, and an independent audit mechanism.
Keywords: algorithmic governance, administrative law, artificial intelligence, DPDP Act 2023, Aadhaar, faceless assessment, facial recognition, accountability.
