Understanding the Viability of Artificial Intelligence from the Lens of Judicial Discretion

Artificial Intelligence (AI) though nascent has become pervasive in the mundane bringing it in confrontation with different facets of law and law-making. While the former requires a jurisprudential analysis for evaluating AI as a subject of law, the latter poses challenge to the established State set-up and its three branches of legislature, executive and judiciary. Furthermore, in the latter there arises implementational challenge making the discussion on AI as a tool in law making quite immanent to the legal domain. Those speaking in favour of AI as law-making tool vouch for its assistance to procedural aspects; while those contending against, warn of its mechanical approach devoid of judicial discretion. Judicial discretion, especially within the common law framework, protects against mechanical implementation of law by ensuring for checks and balance. The various tenets of judicial discretion enables the judge and other judicial stakeholders, to engrain the legislations in such a manner that is in conformity with the societal demands. As Dworkin asserts, “The professional opinion of which I speak uses 'discretion' in a more characteristic sense to mean that the judge must sometimes reach his decision by means other than the application of standards, that such standards sometimes leave him free to choose.” Judicial discretion has provided the space for evolution of purposive doctrines enabling efficient administration while balancing it with rights. However, the use of AI as an adjudicatory tool may threaten this rationality by reserving the scope and implementation of law. AI works on the LLM (large language model) such that it learns from the sources around making it more human-like and not humane. It is analytical, logical but does not possess the individualised acumen of reasonable person and therefore is devoid of discretion. Discretion is rooted in the principles of natural justice and therefore mandates a reasoned approach. While AI learns from the judgments and judicial acts, it shall be at the risk of fallacy, being confined to the existing decisions and interpretations impeding the principle of reasoned decision. Furthermore, in inquisitorial systems involvement of AI as a judicial stakeholder poses the risk of altering the procedural aspects. While the substance provides the remedy, it is the procedure that augments it. This paper aims to understand the limitations of AI as a tool in judicial decision-making from the lens of judicial discretion. It delves in the theoretical premise proposed by of Dworkin, Raz, Perry, Greenawalt, Sartorius, Steiner, Posner and others for understanding significance of judicial discretion in balancing the trifurcate State set-up and thereafter discusses the constraints AI may pose to this balance.