Algorithmic Decision-Making And Administrative Discretion In India: From Formal Authority To Substantive Control
Sheikh Abbas Bin Mohd, Research Assistant, Falcon Legal Chambers
Musrat Mohd, Managing Partner, Falcon Legal Chambers
Jaffar Ahmad Khan, Senior Associate, Falcon Legal Chambers
ABSTRACT
Algorithmic systems increasingly participate in public administration by allocating cases, identifying risk, verifying eligibility and structuring the information on which officials act. Scholarship has established that automated decision-making can redistribute discretion, obscure decisional grounds and make nominal human oversight ineffective. This article addresses a narrower problem of legal proof: when an algorithm materially shapes an administrative outcome, what must the administrative record show before a reviewing court can conclude that the statutory decision-maker genuinely exercised the discretion entrusted to that office? Using doctrinal- conceptual analysis of Indian public law, the article synthesises Supreme Court jurisprudence on acting under dictation, non-fettering, application of mind, relevant considerations and reasoned orders. Three documented public-sector applications-Samagra Vedika, Project ADVAIT/GST analytics and faceless income-tax assessment-are used illustratively to test the resulting framework across different degrees of algorithmic influence. The article develops a five-part Substantive Discretion Test (SDT): decisional provenance, comprehension of decisional grounds, effective override, individualised consideration and reason ownership. The SDT is not a new standard of review. It is an evidentiary framework for identifying whether apparent human authority corresponds to substantive exercise of statutory judgment, and for connecting the resulting record to established grounds of judicial review.
Keywords: Algorithmic decision-making; administrative discretion; artificial intelligence; judicial review; human oversight; administrative law; India.
