ABSTRACT
Human use of machine-driven artificial intelligence (AI) has attracted deserved attention. But at the same time, scholars studying the impact of AI have largely overlooked the degree to which humans themselves frequently engage in robotic analysis of legal disputes that can operate in the nature of a failed algorithm. The heavy, sometimes determinative, use by courts of dictionaries and canons of construction as templates for interpretation can be uncomfortably close to decision by software.
This disturbing similarity of machine-like ‘analysis’ by courts may offer some shortcuts to decision but does so at the cost of stripping away nuanced analysis and human judgment. The current dominance of unduly dictionary-centric textualism regarding statutes, regulations, contracts, and insurance policies reflects not only an impoverished approach to interpretation but also the insidious invasion of de facto, defective AI into legal analysis. We make the observation that textualism – and to some extent legal formalism as a whole – is a form of decision making by de facto AI – something insufficiently appreciated by observers. Both application and analysis of textualism and formalism has proceeded without sufficient appreciation that it is a form of AI.
Although deploying some AI-like shortcuts of the human mind as heuristics may to some extent be a necessary evil for efficient functioning of the legal-governmental-commercial system, it often produces sub-optimal results by imposing a reductionist system of undue formalism upon a legal system and society generally better served by a more flexible, holistic, functional approach to resolving issues of law and policy. At a minimum, judges, regulators, and policymakers should appreciate the extent to which they often let dictionary definitions and canons of construction become an example of outsourcing thinking to tools that do not engage in any actual analysis. We posit that jurisprudence is better served by the presence of a more flexible, less formulaic analysis that is cautious in its use of dictionaries even if formalism/textualism remains the presumptive approach to construction of insurance policies and other texts.
Knutsen, Erik S and Stempel, Jeffrey W, Textual Literalism, Formalism, and Canon-Centric Construction as De Facto – and Defective – Artificial Intelligence (January 19, 2022) in Abel Viego Copo and Miguel Martinez Munoz, eds, Proceedings: III Congreso International de Derecho Del Seguo, III International Conference on Insurance Law (Reuters, 2022).
Leave a Reply