

Access to AI tools alone cannot guarantee access to justice as technology cannot replace the human judgment required to decide cases, Senior Advocate Rajshekhar Rao said on Saturday.
Rao said AI could improve court efficiency, widen access to justice and reduce the burden of routine legal work, but the responsibility of deciding cases must remain with human judges.
“We must keep in mind, access to a chatbot is not access to justice. Access to a translated document does not necessarily mean that the person has the right translation. A word misunderstood by the chatbot alters the entire basis of the judgment. That is why we have this whole aspect of hallucination,” said Rao.
He was speaking at the inaugural JusTech National Conclave at Dharmashastra National Law University (DNLU), Jabalpur, on the use of artificial intelligence and technology in the Indian judiciary.
The first session, titled “Algorithmic Justice in the Driver’s Seat: Reincarnating the Judiciary through AI and Technology”, featured Supreme Court Justice Joymalya Bagchi, Madhya Pradesh High Court judge Justice Anand Pathak, Senior Advocates Rajshekhar Rao and S Gurukrishna Kumar, and Advocate Siddharth R Gupta.
Rao said that it is the judge who should be firmly in the driver's seat when justice is being administered.
“The question is not whether artificial intelligence will transform the judiciary. It almost certainly will. The more consequential question, to my mind, is who or what, as the topic suggests, should be in the driver's seat when justice is being administered? On that again, I believe there is no doubt. It is the judge,” he said.
He stressed that adjudication involved more than processing information, as judges have to assess evidence, interpret laws and exercise discretion while deciding cases.
“Adjudication is not simply the processing of information. It involves the evaluation of competing claims, assessment of evidence, interpretation of legal principles, the application of law to facts and, more importantly, the exercise of judicial discretion,” said Rao.
Additionally, he also cautioned lawyers against blindly relying on AI-generated research and summaries. He held that technological competence without legal judgment could be dangerous.
Senior Advocate S Gurukrishna Kumar flagged the risks of bias in AI systems, citing an Amazon recruitment tool that reportedly favoured male candidates because it had been trained largely on resumes submitted by men.
“The quality of the input determines the quality of the output. So the first step, therefore, is input being the determinative factor. You have to necessarily look at AI as a tool,” he said.
Kumar said AI could help courts prioritise cases based on pendency, assist in deciding which cases required urgent attention and summarise voluminous evidence. However, judicial discretion must remain with human judges.
He also cited instances of AI systems producing erroneous results, including a traffic challan issued after a camera-based system mistook a guitar carried by a motorist for a pillion rider without a helmet.
Kumar stressed the need for human verification and ethical safeguards before AI-generated outputs were relied upon.
He also warned that indiscriminate adoption of AI could affect employment and human dignity.
“AI is a wonderful servant but can be a dangerous master. Let us adopt them. Let us not put them in the driver's seat. Let us have them as a tool and let us be in the driver's seat. Algorithmic justice, yes, by all means, but with human intervention,” said Kumar.
Advocate Siddharth R Gupta referred to a Supreme Court judgment dealing with AI-generated fake citations and said the Court had compared AI hallucinations to methyl isocyanide, the toxic gas responsible for the Bhopal gas disaster.
He recounted an instance in which a non-existent judgment titled Mercy vs. Mankind was cited during arguments before the Supreme Court. The fake citation was generated by AI.
“It is so misleading that it defeats the very purpose for which technology and AI have been brought into the system. And I feel we should have zero tolerance towards it,” said Gupta.
Gupta, however, said AI could help courts through machine learning, natural language processing, predictive analysis and automated administrative systems. He cited Estonia's AI judge pilot for disputes valued below 7,000 euros and AI-based adjudication systems in Singapore and China for certain categories of disputes.
He cautioned, however, against allowing machines to take over judicial decision-making.
“Howsoever convenient technology may be, it should not substitute the attributes of human reasoning. We may reincarnate the judiciary with AI, but it should be AI-assisted adjudication, not AI-driven adjudication,” said Gupta.
[Read Live Coverage]