ALGORITHMIC ACCOUNTABILITY IN PAKISTAN: CONSTITUTIONAL CHALLENGES TO AUTOMATED DECISION-MAKING AND THE PRESERVATION OF DUE PROCESS

Abstract

Pakistan’s rapid deployment of automated decision-making (ADM) architectures, from NADRA’s biometric vetting and FBR’s risk-profiling engines to Safe-City predictive-policing platforms, has shifted administrative discretion from public officials to opaque statistical models. This research article interrogates, through a qualitative doctrinal method, whether such deployments withstand the procedural safeguards embedded in Article 10-A of the Constitution, the General Clauses Act 1897 s 24-A, and allied jurisprudence that prizes “reasoned orders” over inscrutable outputs. Drawing upon newly compiled case studies, field interviews, and forensic code audits, we expose three systemic deficits: first, algorithmic opacity frustrates the judicial requirement for intelligible reasons; second, liability for erroneous machine judgements remains unassigned within PECA 2016 and successive Data-Protection Bills; and third, the resultant evidentiary vacuum dilutes courts’ power of judicial review and threatens the rule-of-law architecture. To redress these deficits, we formulate a context-sensitive accountability standard that couples ex-ante algorithmic impact assessments with an ex-post “right to explanation,” harmonizing Pakistani administrative law with comparative best practices under the GDPR. We further propose a statutory mandate for an independent Algorithmic Audit Commission equipped to certify explainable AI systems prior to public procurement. Our findings demonstrate that constitutional due process in the twenty-first century is inseparable from algorithmic process, and that without transparency, efficiency gains purchased through ADM will remain constitutionally, deeply, and democratically infirm.