AI for judiciary will help reduce pendency of cases

SUVAS, a machine learning tool to process language, is already being used to translate SC judgments into regional languages.

There are nearly 4.4 crore cases pending in Indian courts at all levels—the Supreme Court, the high courts and the district/taluka courts. The bulk of these cases have been pending for more than a year. As FE has pointed out earlier, there are many underlying factors, from chronic shortage of judges to infrastructure gaps. But, fixing these alone won’t help, as the Economic Survey 2018-19 showed, because clearing pending cases by 2024 would need going beyond merely meeting the sanctioned strength in lower courts, already nearly 20% short of the sanctioned number of judges. The pandemic making in-person adjudication of cases in the ‘business as usual’ manner a significant transmission risk only exacerbates this problem. Thus, solutions that expedite litigatory processes and labour by increasing productivity, like deploying artificial intelligence (AI), have become all the more vital. India has already made a start, rolling out SUVAS (Supreme Court Vidhik Anuvaad Software) in November 2019 and SUPACE (Supreme Court Portal for Assistance in Courts Efficiency) earlier this month.

SUVAS, a machine learning tool to process language, is already being used to translate SC judgments into regional languages. With quick access to case law as set by SC judgments, in a comprehensible-for-all format, it is likely that court processes in the high courts and lower courts could get accelerated. SUPACE, a composite AI-assisted tool, can be used to push up efficiency of legal researchers and judges; it will read case files, extract relevant information, draft case documents and manage apportioning of work. If facts and arguments relevant to judging a particular case are intelligently presented in a matter of seconds—done manually, this would have taken months—adjudication could become that much faster. As per a report on indiaai.gov.in, SUPACE is customisable, that is, it can behave uniquely like an individual user, learning from and mirroring user behaviour; to illustrate, imagine a system that learns to glean relevant data and present it in a structure that a judge/legal researcher finds easy to comprehend or present. As it is with all AI, as the system ‘learns’, efficiency leaps exponentially. The SUPACE system also includes a chatbot that can give the overview of a case, respond to elementary questions, while switching between documents and prompting further questions to sharpen the user’s understanding of a matter; the logic-gate the chatbot relies on constantly picks up bits of knowledge, thereby refining responses to both factual and contextual questions.

While the Supreme Court has directed the high courts to start using the system, the deployment of AI would also raise many questions on the reliability of such systems and the chances of human biases being carried forward or even getting amplified. To that end, AI researchers will face some knotty questions, but nothing that can’t get resolved with the march of technology. To engender trust, wide consultative processes need to be undertaken. Transparency on AI systems’ selection of data/information, and analysis of this, would be vital.