AI in India's Courts: What Justice Manmohan's 60% figure means for legal work
At the India Law AI and Tech Summit 2025 in Delhi, Supreme Court Justice Manmohan said artificial intelligence could resolve over 60% of India's pending litigation by automating routine, small-ticket matters. Think traffic challans, cheque-bounce cases, and minor offenses. These disputes demand minimal judicial reasoning yet clog listings and staff time. Freeing that bandwidth would let courts focus on the harder 40% that actually need deep legal analysis.
What can be automated without bending the law
Use AI to triage, draft, and schedule repetitive matters that follow predictable patterns. Bulk filings, standard notices, cause-list prep, and first-level scrutiny are ripe for automation. Justice Manmohan also pointed to clustering: if thousands of disputes raise the same issues (e.g., land acquisition), a single ruling can dispose of a large batch efficiently.
SU-PACE: Helpful assistant, not a judge
The Supreme Court has piloted SU-PACE, a "digital research assistant" that reads case files, extracts issues, summarizes content, and flags relevant precedents. It does not deliver judgments. The intent is clear: a hybrid model where tech handles clerical load while judges focus on reasoning, fairness, and remedies. The bench remains the final gatekeeper.
Real risks that need real guardrails
- Hallucinated case law: AI can fabricate citations or misstate holdings. Every citation needs verification.
- Algorithmic bias: Training data can skew recommendations; periodic audits are non-negotiable.
- Privacy and confidentiality: Case materials must be protected with strict access controls and clear data retention policies.
Justice Manmohan underscored all three. Scale should wait until safeguards are in place and consistently tested.
The warning from the Bar
Dr Lalit Bhasin, President of the Society of Indian Law Firms, cautioned against over-reliance. In interviews, young lawyers were turning in near-identical AI-drafted work-signs of declining critical thought. The message: use AI as a tool, keep the judgment human.
The pendency problem won't fix itself
Pendency has crossed 5 crore cases as of 2025, up from around 5 crore in 2023-24, according to public dashboards. That drag isn't only about speed-it's about statutory bloat: excessive, outdated, overlapping, and poorly drafted laws. Structural clean-up must run in parallel with technology or delays will persist.
National Judicial Data Grid numbers make the scale plain, and the Supreme Court of India has already begun testing support tools. The next step is disciplined rollout with measurable outcomes.
Practical next steps for legal teams
- Start with a "low-risk, high-volume" list: traffic challans, cheque-bounce matters, petty offenses, routine interlocutory applications.
- Standardize inputs: build checklists, templates, and data schemas so AI systems have clean, consistent fields.
- Adopt a two-layer review: machine draft, human verify-especially citations, ratios, and statutory references.
- Create an AI governance note: sources used, audit cadence, redlines for privacy, and escalation paths for errors.
- Batch similar cases: use clustering to prep one master brief or order outline that can serve a large group with minimal tweaks.
- Track impact: measure cycle time, error rates, and listing-to-disposal intervals before and after deployment.
- Train juniors on "AI with reasoning": prompts are fine, but insist on manual issue-spotting and ratio verification.
Where this is headed
If courts automate the routine and clean up dockets in bulk, the remaining matters will finally get the attention they deserve. The bar's role will shift: fewer hours on clerical tasks, more time on strategy, evidence, and settlement. That's healthier for clients and the system-provided integrity and due process stay non-negotiable.
Skill up without losing the plot
Lawyers who can supervise AI, verify outputs, and translate results into courtroom-ready arguments will stand out. If you're building that capability across your team, explore curated options here: AI courses by job.
Your membership also unlocks: