Artificial Intelligence in Pakistan’s Courts: Legal Accountability and Due Process

Abstract

ABSTRACT

There are good reasons for the courts of Pakistan to be interested in artificial intelligence (AI), which has potential benefits for legal research, case management, translation and the analysis of judicial materials. There will be some 2.36 million cases pending at the end of 2024. But the use of AI in adjudication also raises concerns about transparency, bias, privacy, accuracy and accountability. This article aims to examine the legal parameters within which the integration of AI-assisted systems into the judicial institutions of Pakistan can be constitutionally consistent and procedurally fair. It distinguishes between administrative applications of AI and systems impacting substantive judicial decisions, by way of doctrinal analysis of Articles 4, 10A, 14, 19A and 25 of the Constitution, the Qanun-e-Shahadat Order, 1984, the procedural obligation to provide reasons, and the Supreme Court’s 2025 judgement in Ishfaq Ahmed v. Mushtaq Ahmed, along with a comparative study of governance models in the European Union, England and Wales, Canada, India and Colombia. It argues that the constitutional problem is not the use of AI per se but the risk that unexamined machine outputs are fed into the reasoning of a decision without the parties being aware of it, and without the judge subjecting it to genuine scrutiny. The article finds that the existing law in Pakistan already contains the principles needed to govern judicial AI, but lacks the rules, data-protection legislation and institutional capacity to apply them. Suggests a proportionate, risk-based governance framework based on mandatory human review of consequential outputs, testing, audit trails, disclosure, training and effective remedies.