Judicial Integrity in the Age of AI: Addressing the Challenge of Algorithmic Hallucinations in Quasi-Judicial Orders

The Supreme Court’s recent intervention to quash a ₹425.27 crore penalty imposed by a Customs official marks a watershed moment in the intersection of technology and administrative law. The order was found to be fundamentally flawed, as it relied on 'AI hallucinations'—fabricated case laws and non-existent legal citations generated by artificial intelligence tools. This incident highlights a growing concern regarding the uncritical adoption of generative AI in public administration and quasi-judicial processes. While AI offers the potential to enhance efficiency, reduce pendency, and assist in complex data analysis, it cannot serve as a substitute for the human application of mind. The core of judicial and quasi-judicial decision-making lies in the exercise of discretion, interpretation of legal principles, and the application of precedent—tasks that require human accountability and ethical judgment. When officials rely on AI without rigorous verification, they risk undermining the rule of law and violating the principles of natural justice. The 'black box' nature of AI models, which can produce plausible but entirely false information, poses a significant threat to the transparency and reliability of government orders.