Using AI In The Courts: Gujarat HC Flags Risks, Backs Human Judgement Over Algorithms
· Free Press Journal
There is considerable expectation that Artificial Intelligence (AI) will contribute significantly to cut the staggering burden of pending court cases, but some High Courts are rightly cautioning against looking at it as a panacea.
The Gujarat High Court recently published a policy that prohibits the use of AI for adjudication, legal reasoning, judicial decision-making, issuing a substantive order, judgment drafting, and bail or sentencing considerations.
Visit saltysenoritaaz.org for more information.
Ultimate responsibility for all orders, judgments and observations will rest with the judge, the court said, and the policy covers all judicial officers, officers, staff and technical personnel of courts at different levels.
The high court amplified well-founded concerns about AI, notwithstanding its useful role in speeding up tasks, given the evolving nature of the technology.
Areas where AI can assist courts
However, there are several areas in which the productivity and efficiency-enhancing capabilities of AI can speed up dispensation of justice, starting with chatbots to interact with those seeking help.
Some of these tasks include legal research support, drafting assistance, language and translation, and case management, which would contribute to case prioritisation and workload metrics analyses.
Where the use of AI — notably generative AI — requires utmost caution is in confidentiality, individual privacy, and scenarios where existing biases would get reinforced and compounded by machines that learn only from existing inputs and may hallucinate.
Gujarat’s precautionary approach mirrors last year’s policy issued by the Kerala HC on the use of AI in district courts. These early initiatives at setting up guardrails for AI use should ideally be followed up with a national consensus led by the Supreme Court.
Balancing technology and human judgement
The estimated pendency of 50 million court cases in India and the incremental nature of administrative reforms understandably raises a demand for a transformation, through faster computerisation and use of AI.
Supreme Court judge Vikram Nath, who released the Gujarat HC policy, correctly underscored the importance of human interpretation over machine algorithms, in areas which require empathy, fairness and a sense of justice.
The Law Society of the UK issued its own advisory to solicitors and their staff on the use of generative AI as a measure of efficiency, but with the caveat on the inability of AI to understand its own output in the way that a human can.
Singapore also places ultimate responsibility on the judicial officer and prohibits disclosure of confidential records to external AI tools. The American Bar Association calls for confidentiality and conflict-checking by lawyers when using AI.
Wherever courts maintain detailed data on legal administration, AI can help identify and fix bottlenecks and undue delays. Yet, AI cannot speed up the actual consideration of cases, given the low ratio of high court judges to population, at 1 for 18.7 lakh people, according to the India Justice Report 2025. Solving that needs political will.