Artificial intelligence and ChatGPT in the legal context

In this issue of IJMS, Parikh et al. others have written a timely review titled “Judge Juan Manuel Padilla Garcia, Chat Generative Pre-Trained Transformer (ChatGPT) and a controversial medicolegal milestone.”^[1] They have outlined the first case in the world, where a judge has specifically documented the use of OpenAIs ChatGPT in his order. His court in Colombia was hearing a case regarding who should bear the cost of the treatment of an autistic minor. ChatGPT (launched on November 30, 2022) has already taken our world by storm. A simple Google search threw up 629 million hits on February 19, 2023.^[2]

The Reserve Bank of India brought out a discussion paper on the application of artificial intelligence (AI) in the financial sector.^[3] Similarly, the Ministry of Health and Family Welfare, Government of India produced National Digital Health Blueprint as a regulatory framework guideline for AI in healthcare.^[4] Our Supreme Court has also set up an AI Committee, with a focus on the translation of legal documents; process automation; increasing administrative effectiveness; automating forecasting, prediction and filing; scheduling of cases; and early case resolution using chatbots.^[5] In fact, Supreme Court Chief Justice DY Chandrachud has announced that such technology that promotes efficiency, transparency, and objectivity should become a facilitative tool for judges to facilitate providing greater access to justice for all citizens without any discrimination or bias.^[6]

As Parikh et al. have explained, today’s legal disputes involve millions of documents saved as soft copies. Often they number in tens or hundreds of millions of files. It is therefore impossible to scrutinize them in a systematic, timely, and orderly manner. Under the circumstances, e-discovery is the logical choice using AI, machine learning, and keywords.^[1] However, with computer software, it is common knowledge that garbage in, garbage out. AI will also be only as efficient as the stringency with which it is developed. Any bias in inputs will impact the machine learning process and finally become evident in the system output. Being cognizant of such potential biases, in the case of Mohit Sharma versus the Union of India, the Court had asked the government to ensure that AI is not used in a discriminatory manner against any specific group.^[7]

There is a need for a robust legislative framework for AI before it can be utilized with confidence and reliability in day-to-day practice. Niti Aayog (Govt of India) has put forth a tentative National Strategy for Artificial Intelligence and the State of Telangana had even announced 2020 as the year of AI.^[8,9] As the use of AI becomes more commonplace, businesses and the general public must be aware of its legal implications.

Can ChatGPT (or other similar AI bots) have the ability to create legal documents, conduct legal research, perform ediscovery, and prepare legal opinions? Will this process be transparent and ethical? Who will be accountable for their outputs? ChatGPT has already “passed” the bar examination at a Law University in the USA and qualified as a “lawyer.”^[10]

Hence, contracts, agreements, and pleadings are a few of the legal documents that ChatGPT can produce in a matter of minutes. It will save a lot of time for lawyers and might even make many of their services redundant. ChatGPT can also evaluate human produced legal documents to identify (and help correct) mistakes and discrepancies in draft legal documents. Whatever (or whoever) may have produced the final version of the legal document, a contract to be enforceable must be fully understood and entered freely. Hence, a human lawyer will have to be involved in explaining the implications to a human client (details in R. Kalyani versus Janak C. Mehta).^[11]

In the case of State of Maharashtra versus Dr. Praful B. Desai, the courts have stated that “judicial discretion cannot be a matter of irrationality or caprice, but must be guided by the law and the precedents that have been created over centuries of legal history.”^[12] Clearly, legal decision-making cannot be left to AI bots like ChatGPT and AI cannot replace human judges. Moreover, the reference to ChatGPT in the order written by Judge Garcia is a controversial decision that defies logic.^[1] Especially since ChatGPT has been around for only a few months; has been shown to give contradictory/different answers to the same question; and can only refer to its limited database (items known on or before the year 2021). Legal judgments are required to be based on uncompromised standards of justice, fairness, morality and human dignity instead of following arbitrary or subjective standards as mentioned in the case of Navtej Singh Johar versus the Union of India.^[13]

When asked, ChatGPT is prompt in admitting its limitation while being used in the judiciary. It says, “it cannot be a substitute for the knowledge, expertise, and logical thinking process of a human required to give a moral judgment in the right context”.^[1]

History tells us that AI is set to permeate our lives in an unprecedented manner. Whether it is ChatGPT or its cousins (Googles Bard), bots will become an invaluable component of the legal system during our lifetimes.