Data and justice: On courts in India and AI tools

In the latest step in a long-standing effort to digitise the judiciary, Chief Justice of India (CJI) Surya Kant announced two initiatives from the Bench, called ‘One Case, One Data’ (OCOD), a unified judicial data platform, and ‘Su-Sahayak’, an AI-powered chatbot on the Supreme Court of India website. OCOD promises a unified digital trail for a dispute as it moves through various courts, linkages between court records and litigant actions (such as appeals), easier access to various documents, lower need for manual verification, reciprocal access to High Courts and other courts, and more accurate judicial statistics. It is notable considering the wide variation in software practices and records quality across India’s thousands of district and subordinate courts. If the programme succeeds, standardised data could also allow administrators to determine where cases are held up and ease procedural bottlenecks, and improve data-based decision-making overall. ‘Su-Sahayak’ has been integrated into the Court website’s front-end to help users navigate case status, cause lists, orders and judgments, e-services, and frequently asked questions. As with any major state-backed technology rollout in India, questions about interoperability, integrity of legacy records, restricting access to private information, and staff skilling remain. By aspiring to a centralised digital fingerprint for each case, OCOD also bears the risk of misuse.

While the CJI said that these new tools will improve “access to justice”, their introduction risks deepening the digital divide. OCOD may require lawyers to maintain digital scanners, cloud backup options, and updated software. Metropolitan corporate firms can easily absorb these costs but independent practitioners in the district and taluka levels will lack the capital. The system may also introduce digital middlemen to help litigants who cannot navigate the e-filing portal, potentially creating a new layer of unregulated costs. While the government has launched assistants with voice-first capabilities, such as Jan Sahayak, ‘Su-Sahayak’ is primarily text-based and could exclude people who are not comfortable typing or navigating complex website menus. The state and the judiciary must ensure that the AI model is not biased against marginalised communities who were historically disproportionately arrested or denied bail. India’s courts have generally been more comfortable with AI for assistance than for substantive reasoning. ‘Su-Sahayak’ follows SUVAS, to translate judgments, and SUPACE, which processes facts and legal precedents. The line must continue to hold as the judiciary adopts more powerful tools that have already tempted practitioners in other domains to abuse them.