AI in legal practice still has many hoops to jump through

Judges, lawyers, law firms and even cops hope to leverage Artificial Intelligence capabilities to generate and extract relevant information to improve their efficiency.

The legal practice in India is anachronistic, colonial, and in urgent need of modernisation to increase efficiency and operational capabilities. The legal practice involves multifarious tasks, each having its own set of complexities. Judges and counsel appearing for the parties to a legal dispute are required to sift through voluminous pleadings and documents, plethora of statutes and innumerable judgments.

All this requires in-depth research in their respective roles by lawyers and judges. Meanwhile, almost five crore cases are pending in India, emblematic of how litigious a country we are, and the need for greater efficiencies.

Extractive And Generative AI

This is where Artificial Intelligence (AI) is poised to revolutionise the legal industry. Two types of AI technologies – Extractive and Generative AIs – are set to completely reform the way law firms and lawyers operate.

Extractive AI software can be used to extract relevant pieces of information from voluminous legal documents and pleadings to answer specific legal questions. This would help lawyers to prepare for their matters with expediency and efficiency.

Generative AI, like ChatGPT, provides assistance in drafting the basic framework of a legal document. Such software can be deployed to create legal documents of a skeletal nature which can then be embellished as per the specific requirements of a particular case. Generative AI, thus, would save time for lawyers, concerned would no longer be required to prepare a legal document from  scratch.

Harvey AI, a generative software developed on OpenAI’s large learning model GPT-4, has been deployed by several law firms. The software enables lawyers to enter complex legal tasks in simple instructions to achieve desired results.

India Makes A Start

The Supreme Court of India has also begun using AI software to transcribe arguments before the court and dialogue between the bar and the bench. This has made the apex court a court of record in the true sense. The court has also constituted a committee on AI to study the technology and augment the justice delivery system.

The Kerala high court has started using a machine learning software from July 10 to identify defects in bail applications. This technology would reduce human intervention and streamline the process of e-filing.

In a first, Punjab and Haryana High Court's Justice Anoop Chitkara employed ChatGPT to assist him in authoring a bail order. The court was dealing with a bail application in an assault case. The court framed its question: "What is the jurisprudence on bail when the assailants have assaulted with cruelty?" - to which the software gave a detailed opinion.

The court, however, clarified that ChatGPT’s response must not be construed as the court's own observation. But this endeavour is a stark reminder that not only lawyers require AI to enhance their operational capabilities, judges also need assistive technologies for better administration of justice.

Improving Justice Delivery

Any pleading filed in a court, has to, at the very outset, pass through the scrutiny of the registry for it to be listed before a judge. Process of scrutiny and curing of defects ends up in needless delays and squandering of human resources. AI eases tasks such as scrutiny of pleadings, minimising human intervention and ensuring timely listing before courts.

Indian appellate judges are assisted by legal researchers and judicial clerks who are entrusted with preparing one-pager notes by condensing all the particulars of a case in a concise manner. Judges read these notes to understand the gist of a case. Legal clerks also assist judges in writing judgments, orders, and legal research.

Extractive AI technologies can obtain summaries of cases and save precious judicial time, which can then be channelled for legal research and judgment writing. This would ultimately lead to speedier justice.

Similarly, in the criminal justice system that is based on the cardinal principle of “innocent until proven guilty” – it is paramount that red tape does not prolong incarceration of accused persons. The use of machine learning and AI technologies must be explored in generating chargesheets for the police, on feeding with the FIR, statements of witnesses, forensic evidence, etc, followed by extensive human supervision to scrutinise for algorithm bias. This would oil the wheels of the criminal justice system, help spot deficiencies in chargesheets, while not leaving everything to AI.

Operationalising AI Not Easy

The behemoth that AI is also comes with its own distinct drawbacks. The risks of predictive policing like racial bias are well known. In June 2023, a New York-based lawyer had to apologise to a judge for submitting a case brief riddled with fabricated judicial precedents and legal opinions that AI had concocted.

AI poses novel problems too. For instance, while drafting a pleading, AI might omit to include special facts of a case like psychological positioning of client or grounds of sympathy which are only privy to the lawyer concerned.

Moreover, while some judges rely more on written submissions, others are inclined towards oral arguments. In such peculiar scenarios, AI will find it hard to draft pleadings according to the judicial temperament and psyche of a particular judge.

Although AI can revolutionise the legal industry by making processes smooth, it is still at a very nascent stage for us to pass a judgment on the extent of its utility and success rate in operational scenarios i.e. a courtroom.

AI use can indeed provide viable assistance to legal practitioners and even judges but it cannot, in any manner, replace human beings. The human mind, undoubtedly, remains the most advanced and intelligent machine which is able to apply its time-honoured experience to solve complex legal problems. Errors and hallucinations by AI software presently allay apprehensions that AI will replace human beings and thereby reduce employment opportunities.

Kaustubh Mehta is an advocate who practises in Delhi. Views are personal, and do not represent the stand of this publication.