AI is flooding courts – but it’s not changing who wins

However, despite this surge in adoption, Nandan Subramaniam, the principal of Zed Law, has stressed that it has yet to meaningfully influence litigation outcomes.

Why AI use in court filings has taken off

The rapid adoption of AI in legal drafting has been driven by the convergence of several forces rather than a single breakthrough, with Subramaniam pointing to three developments that all “landed at once” to reshape the profession.

One of the key drivers he identified was the technology reaching a level of reliability and trustworthiness at a time when lawyers were increasingly looking to move away from routine research and focus instead on more substantive legal thinking.

“The tools got good enough to trust with real work, they got cheap, and they arrived in a profession drowning in low-value time,” he said.

“Lawyers don’t want to spend the day on research and wading through copious documentation. That was never the part that made the work valuable.

VIEW ALL

“AI lets you get to the drafting, the actual thinking, and skip the hard yakka. Once that was possible, adoption was always going to move fast.”

Beyond improving efficiency for legal practitioners, Subramaniam said AI is also reshaping access to justice by enabling self-represented litigants to prepare legal filings in hours instead of weeks and pursue claims that might otherwise never have seen the light of day.

“The other half is access. AI has put legal drafting in the hands of people who could never afford a lawyer. A filing that used to take a self-represented person weeks, or never got made at all, now takes an afternoon,” he said.

Against this backdrop, he argued the surge in AI-generated filings reflects more than just technological progress, signalling a legal system being reshaped by both greater productivity within the profession and broader access to justice for those previously locked out of it.

“So the [high number of AI-drafted court filings] isn’t one story,” he said.

“It’s lawyers cutting the grunt work, and it’s a whole population getting into the system who were shut out before. Both are pushing the number up at once.”

So why isn’t AI changing who wins in court?

Despite the dramatic rise of AI-assisted drafting and its growing presence in court filings, Subramaniam argued that the technology is not changing the one metric that matters most: who ultimately wins in court.

But why is this happening?

Subramaniam pointed out that while AI can help lawyers draft faster and more effectively, it can’t alter the underlying facts of a case or the way the law applies to them – the two factors that ultimately determine the outcome of litigation.

“The win rate hasn’t moved because cases aren’t won on drafting. They’re won on the law applied to the facts, and AI changes neither,” he said.

“The facts are what they are. A judge finds them. A cleaner, better-argued version of a weak case is still a weak case.”

While AI can produce polished legal writing in seconds, he warned it still lacks the legal judgement required for litigation, frequently producing outputs that sound authoritative but unravel when tested against procedural requirements.

“What AI is good at is the writing. What it can’t do is the judgement underneath. It doesn’t know what’s admissible, what to plead, what to leave out, or when to settle,” he said.

“It’ll hand you a confident answer that’s procedurally hopeless.”

For all its strengths, Subramaniam cautioned that AI remains vulnerable where litigation is least forgiving, with a single hallucinated authority or procedural mistake capable of derailing an otherwise strong case.

“Court rules and practice directions are unforgiving, and that’s where AI still falls down,” he said.

“The only way it moves an outcome is the wrong way. Feed in a hallucinated case, miss it on the review, and you lose a matter you should have won and be extremely embarrassed.”

What AI is really doing inside the legal system

Rather than transforming litigation outcomes, Subramaniam pointed out that AI’s true disruption lies in reshaping not only how legal work is produced but also who now has the power to produce it.

“AI has become very good at generating the document. It’s done nothing to supply the things that decide a matter, which are strategy, judgement and accountability,” he said.

He argued that AI’s most significant contribution is its ability to dramatically reduce the time and cost of drafting, placing the same productivity gains in the hands of senior counsel and self-represented litigants alike.

“Its real role is to compress the cost and time of getting words on a page, at both ends of the market, for the silk and the self-represented litigant alike,” he said.

“That tells you something uncomfortable about legal work.”

While technology can now handle much of the administrative heavy lifting, Subramaniam stressed that the true value lawyers bring still lies in judgement, strategic decision making, and accountability when the stakes are highest.

“The parts AI has swallowed were always the commodity parts. Research, first drafts, summaries,” he said.

“The parts it can’t touch are the parts that were always the point, knowing what matters, reading the other side, carrying the risk of being wrong.”

What happens next?

Looking ahead, Subramaniam said he sees no realistic scenario in which AI adoption slows, arguing that the technology is advancing at such a rapid pace while the cost of access continues to fall, making wider uptake all but inevitable.

“Yes, and there’s no version where it reverses,” he said.

“The tools improve every quarter and the barrier to using them keeps dropping. AI-assisted drafting and research is the default now, not the exception, whether you’re top-tier or lodging a claim from your kitchen table.”

While AI adoption is expected to continue accelerating, Subramaniam said the focus in the years ahead should not be how widely the technology is adopted, but whether it will have any meaningful impact on litigation outcomes.

“What I’d watch isn’t the volume, that’s a given. It’s whether the win rate ever moves, and I don’t think it will by much, because the things that decide cases still aren’t the things AI is good at,” he said.

Subramaniam argued the next chapter of AI in litigation will be defined less by technological breakthroughs than by the courts’ growing scrutiny of how the technology is used, even as its presence in legal disputes continues to expand.

“The more interesting growth is in the rules. Disclosure obligations, hallucinated authorities, who carries the liability when the advice is wrong, whether your chats with a chatbot are privileged,” he said.

“Courts here have already started putting guidance around AI use, and that tightens from here; it doesn’t loosen.

“So expect more AI, more filings, more of it in litigation. Just don’t expect it to change who wins.”