Every few months a headline pops up claiming that an “AI judge” has started deciding cases somewhere in India. It makes for a great thumbnail. It is also not how things work, at least not yet. What is real is quieter and, honestly, more interesting: Indian courts have been bringing AI into the parts of the justice system that eat up the most time, while keeping the actual judging firmly in human hands.
Here is a plain-language look at what AI is doing in Indian courts today, where it could genuinely help, and the hard questions nobody should skip.
Why courts are looking at AI in the first place
The short answer is the backlog. According to the National Judicial Data Grid, cases pending across Indian courts run into crores, and a large share of them sit in district courts for years. Most of that delay is not about judges taking too long to think. It comes from paperwork: translating records, finding the right precedents, scheduling, tracking adjournments and retrieving files.
That is exactly the kind of repetitive, document-heavy work where software is good and people get worn out. So when you hear “AI in courts”, think research assistant and filing clerk, not robot on the bench.
What is actually in use
SUPACE: a research assistant for judges
In 2021 the Supreme Court launched SUPACE (Supreme Court Portal for Assistance in Court Efficiency). It helps judges and their staff go through large case files, pull out relevant facts and find related material faster. It does not suggest a verdict. The court was clear about that when it was introduced.
SUVAS and translation tools
Judgments are mostly written in English, while litigants across the country speak dozens of languages. Machine translation tools such as SUVAS have been used to translate judgments into regional languages so that people can actually read the decisions that affect them. Translations are still checked by humans before they are relied on.
eCourts Phase III
The government’s eCourts project is now in its third phase, focused on digitising records, e-filing, virtual hearings and smarter case management. AI features such as automatic categorisation of cases, transcription and data analysis fit naturally into this plan, because you cannot apply AI to files that only exist on paper.
Could AI ever decide a case?
Technically, software can already produce a draft order for very simple, rule-based matters. Some countries have experimented with automated help for small claims and traffic fines, and China has invested heavily in what it calls “smart courts”. But there is a big difference between a tool that drafts and a system that decides.
In India, any move in that direction would have to deal with a few serious problems first:
- Bias in old data. An AI trained on past judgments learns past patterns, including unfair ones. If certain groups were treated more harshly before, a model can quietly repeat that.
- Explaining the reasoning. A judgment has to show why. Many AI models cannot reliably explain how they reached an answer, and a verdict you cannot explain is a verdict you cannot appeal properly.
- Accountability. If a human judge gets it wrong there is a clear path to challenge it. With a machine, it is not obvious who is responsible: the court, the vendor or the people who trained it.
- Privacy. Court files are full of personal details. Sending them through AI systems raises questions under India’s data protection law that still need clear answers.
- Made-up citations. General-purpose chatbots have been caught inventing case law. Lawyers in several countries have been embarrassed in court for filing AI-written briefs with fake precedents. Any legal AI has to be tightly limited to verified sources.
Where AI can help right now
The most useful near-term gains are not glamorous:
- Sorting new filings by type so urgent matters reach the right bench sooner
- Flagging cases stuck on repeated adjournments
- Transcribing hearings and witness statements
- Translating orders into local languages
- Helping court staff find similar past cases in seconds
- Giving litigants simple updates on where their case stands
None of these replace a judge. All of them give judges more time to do the part only they can do.
What this means for lawyers, litigants and businesses
For lawyers, legal research tools are getting faster and cheaper, and clients will expect quicker turnaround on routine work. The skill that matters more now is judgement: checking what a tool produces instead of trusting it blindly.
For businesses, faster case management could slowly make contract disputes and recovery cases less painful, which matters for everyone from small traders chasing unpaid invoices to startups protecting their IP.
For ordinary litigants, the biggest win would simply be knowing what is happening with their case and getting a date that does not keep moving.
The bottom line
India does not have an AI judge, and that is probably a good thing for now. What it has is a slow, steady shift towards AI-assisted courts, where machines handle the paperwork and humans handle justice. If that balance holds, technology could take a real bite out of the backlog without asking anyone to accept a verdict from a black box.
Keep an eye on how eCourts Phase III rolls out and on any guidelines the judiciary issues for AI use. Those will tell you far more about the future of Indian courts than any viral headline.
