Law Is Made, Not Found: Why the Singularity Stops at the Courthouse Door
A superintelligent machine can transform mathematics (which it is doing). It cannot do the same to law, because law is something people make, not something waiting to be discovered (with all due respect to natural law theorists).
The singularity, and what it can't reach
The "singularity" is the idea that AI will eventually improve itself so fast that it outruns human understanding. On that view, every field built on intelligence gets swallowed: science, engineering, medicine, and, the argument goes, law.
I think that argument gets law wrong. It treats law as a puzzle with a right answer that a smarter mind could find faster. Law isn't that kind of thing. A statute is valid because a legislature enacted it, not because it is correct. A precedent binds because a court with authority decided it, not because it is true.
The clearest way to see this is to compare law with a field where the singularity really could change everything: mathematics.
Mathematics: truth is found
A mathematical theorem was true before anyone proved it. The prime numbers didn't wait for Euclid's permission to be infinite. Mathematicians discover results; they don't decide them.
That makes math the ideal field for a superhuman intelligence. A proof is right or wrong regardless of who wrote it, and today it can even be checked by a machine. Proof assistants like Lean verify every logical step, so a proof from AI is exactly as valid as a proof from a Fields medalist.
The machines have already arrived. In July 2025, models from Google DeepMind and OpenAI each scored 35 of 42 points at the International Mathematical Olympiad, the gold-medal line (TechRepublic). In May 2026, an OpenAI model disproved Erdős's 1946 unit distance conjecture. Nine outside mathematicians checked the proof, and Tim Gowers said he would have accepted it for the Annals of Mathematics without hesitation (The Decoder).
Notice what made that result count. Nobody voted on it. Nobody asked whether the AI had standing to do mathematics. The proof worked, and that settled it. In mathematics, a faster and smarter mind is simply a better mathematician, so a singularity really would change the field from top to bottom. As an aside, I encourage you to listen to Professor Steven Strogatz on the The Daily podcast from October 13, 2026 titled “When A.I. Comes for Your Purpose in Life”). It’s one of the best podcasts I’ve listened to in a while and posits that AI is becoming creative, at least in solving mathematic problems.
Law: authority is made
Law works the other way around. No statute existed before a legislature passed it. No rule of contract was waiting in nature for a judge to uncover. People wrote all of it, and people can rewrite it next session.
That changes what "right" means. A legal answer isn't correct the way a proof is correct. It is authoritative because of who gave it and how. A Supreme Court opinion binds lower courts even if most law professors think it is wrong. A five-to-four decision is just as much the law as a unanimous one. And when the people disagree with the courts, they can amend the statute or, with enough effort, the Constitution.
Legitimacy, not accuracy, is the currency. We accept a verdict because a jury of peers reached it, because the judge was appointed or elected through a process we recognize, and because the losing side had its day in court. Due process is the point, not a step on the way to the point.
A superintelligence could predict how a court will rule. It could draft a statute more clearly than any legislative counsel. What it cannot do is make that statute law. Only an institution people have agreed to be governed by can do that.
What AI will change, and what it won't
None of this means lawyers can ignore AI. The practice of law is changing, rapidly. Research, document review, first drafts and contract analysis are already being automated, and the tools will keep getting better. A smarter machine will be a far better legal assistant.
It may also widen access to justice. Most people who face eviction, debt collection or a custody dispute can't afford a lawyer. Good AI help could change that, which matters more than any efficiency gain at a large firm. Although in practice, this may overload the system to an unmanageable point.
But all of that happens inside the system. The machine helps people work with the law. It doesn't become the source of law. Judges still sign the orders, legislatures still pass the bills, and citizens still decide who holds those offices. A singularity in raw intelligence leaves that structure exactly where it was.
The strongest objection
The best counterargument is that people could simply hand the authority over. A government could let AI decide cases or write rules, and then the machine's output would be law.
This isn't purely hypothetical. In September 2025, Albania named an AI system called Diella as a cabinet "minister" in charge of public procurement (Semafor).
But look at how that happened. A prime minister appointed it, under a constitution, answerable to a parliament and voters. Diella's power is borrowed from human institutions, and those institutions can take it back. Delegating a decision isn't the same as giving up the source of authority. If anything, the example proves the point: even an AI official needs people to make it one.
A second objection comes from natural law: the idea that justice is real and discoverable, like a theorem, so a superintelligence could find the true moral law. Perhaps. But even if it did, that answer would bind no one until a human institution adopted it. History is full of moral truths that became law only when people chose to enact them.
Law stays ours
The singularity, if it comes, could leave mathematicians verifying proofs they could never have found. That is a real change in what the field is.
Law is different because it was never about finding the right answer. It is about how a free people decide to govern themselves, and why the losing side accepts the result. A smarter machine can help with almost every part of that work except the part that makes it law. That part belongs to us, because we made it.