The Automation of Legal Reasoning: Customized AI Techniques for the Patent Field

As Artificial Intelligence and Machine Learning continue to transform numerous aspects of our everyday lives, their role in the legal profession is growing in prominence. A subfield of Al with particular applicability to legal analysis is Natural Language Processing (NLP). NLP deals with computational techniques for processing human languages such as English, making it a natural tool for processing the text of statutes, regulations, judicial decisions, contracts, and other legal instruments. Paradoxically, although state-of-the-art Machine Learning and NLP algorithms are able to learn and act upon patterns too complex for humans to perceive, they nevertheless perform poorly on many cognitive tasks that humans routinely perform effortlessly. This profoundly limits the ability of Al to assist in many forms of legal analysis and legal decision making. This article offers two theses. First, notwithstanding impressive progress on NLP tasks in recent years, the state-of-the-art in NLP will remain unable to perform legal analysis for some time. Second, lawyers, legal scholars, and other domain experts can play an integral role in designing Al software that can partially automate legal analysis, overcoming some of the limitations in NLP capabilities