Beyond The Black Box And The Dual Constitutional Crisis Of Healthcare AI In India
Mohamed Asraf & Reema
ABSTRACT
Clinical medicine is undergoing a quiet but consequential transition: rule- based clinical software, whose logic could be audited line by line, is giving way to generative and deep-learning CDSS whose reasoning even their developers cannot fully reconstruct. This shift produces a dual crisis for Indian health jurisprudence. At the systemic level, it tests the positive obligation to furnish equitable healthcare that the Supreme Court read into Article 21 of the Constitution. At the individual level, it strains the doctrine of informed consent, since a patient cannot meaningfully authorize a diagnostic pathway that no one in the room can explain. This paper argues that Indian law currently addresses these two dimensions in isolation — constitutional health jurisprudence through equality and access-based reasoning, and consent through common-law disclosure standards — without a unifying account of what algorithmic opacity does to both simultaneously. Reading Article 21 and Article 14 jurisprudence alongside the disclosure standard set out in Samira Kohli v. Dr. Prabha Manchanda (2008) and the fiduciary obligations created by the Digital Personal Data Protection Act, 2023, the paper develops legal standards for algorithmic explainability, non-discriminatory health data governance, and statutory consent mechanisms suited to black-box clinical AI, and proposes a stratified consent model and a modified negligence standard to operationalize them.
Keywords: algorithmic opacity, right to health, informed consent, Digital Personal Data Protection Act 2023, explainable AI, medical negligence.
