Jurimetrics: The Icebreaker of Legal AI in Mexican Justice

View on LinkedIn

A precise chisel strike has just cracked the ancient glacier of Mexican justice. It was neither a violent nor a revolutionary blow, but a surgical one: magistrate Juan Jaime González Varas used artificial intelligence to calculate bond amounts and, more importantly, documented it in two theses published today in the Semanario Judicial de la Federación.

Digital registry: 2031009

ARTIFICIAL INTELLIGENCE APPLIED IN JURISDICTIONAL PROCEEDINGS. CONSTITUTES A VALID TOOL FOR CALCULATING THE AMOUNT OF BONDS SET IN AMPARO PROCEEDINGS. https://sjf2.scjn.gob.mx/detalle/tesis/2031009

Digital registry: 2031010

ARTIFICIAL INTELLIGENCE APPLIED IN JURISDICTIONAL PROCEEDINGS. MINIMUM ELEMENTS THAT MUST BE OBSERVED FOR ITS ETHICAL AND RESPONSIBLE USE FROM A HUMAN RIGHTS PERSPECTIVE. https://sjf2.scjn.gob.mx/detalle/tesis/2031010

The crack is small, but cracks in ice have a peculiar characteristic: once they start, they are impossible to stop.

The mathematical elephant in the courtroom

"I saw that many courts and judges were doing it manually. When I say manually, it was with their little sheets — some did it by hand on paper and others had their calculator,"

magistrate Juan Jaime González Varas told me. The image is almost tragicomic: district court judges, required by binding precedent to apply complex mathematical formulas, pulling out pocket calculators or setting amounts by gut feeling.

The result was a judicial lottery: $50,000 pesos here, $100,000 there, with no greater justification than "I think this looks right." As Juan Jaime noted: "Lucky you if your case landed in the first court rather than the second." And all because traditional legal education never contemplated that attorneys would also need to be mathematicians, data analysts, and technologists.

Jurimetrics as a Trojan horse

What is brilliant about this approach is its apparent modesty. He did not try to revolutionize the entire judicial system in one stroke. He chose jurimetrics — that intersection of law, mathematics, and statistics — as the entry point. It is difficult to argue against the use of technology to perform objective calculations. Does it matter whether you used an abacus, a calculator, or GPT-5 if the result is correct and verifiable?

But here is the trick: by validating the use of AI for calculations, the floodgates open for much more. Today it is bonds; tomorrow it will be child support, moral damages calculations, value updates. And eventually, case file summaries, jurisprudence analysis, identification of relevant precedents. The initial chisel strike becomes a network of fractures that will transform the entire system.

The four pillars to keep from falling off the cliff

Aware that he was walking on thin ice, the magistrate established four minimum principles — an exercise in "self-restraint," as he calls it:

Proportionality and harmlessness : AI only for what is necessary, without encroaching on the core decision-making function Personal data protection : Never compromise sensitive information Transparency and explainability : Show exactly what was done, how, and why Human oversight : AI as an aid, never as a judge

1. Proportionality and harmlessness : AI only for what is necessary, without encroaching on the core decision-making function 2. Personal data protection : Never compromise sensitive information 3. Transparency and explainability : Show exactly what was done, how, and why 4. Human oversight : AI as an aid, never as a judge

"I ran the model across three different ones," he admitted with striking candor. "I realized that one did not use the current national consumer price index." Rather than concealing these imperfections, he documented them. Because the point is not to feign perfection, but to build trust through transparency.

The open secret no one wanted to admit

"I am certain there are judges and magistrates who are indirectly using artificial intelligence and no one is being transparent about it," Juan Jaime said bluntly. And we all know he is right. ChatGPT is already drafting briefs, Claude is reviewing arguments, Perplexity is searching for precedents. The difference is that now there is a legal framework, a published thesis, an official precedent.

"My idea is for them to copy this [prompt]. To copy this one," the magistrate insists. He is not trying to be the sole innovator; he wants to democratize innovation. So that the district court judge in Chiapas can perform the same objective calculations as the one in Polanco.

The slow-motion tsunami

What we are witnessing today is not merely the adoption of a technological tool. It is the beginning of a transformation that Mexican justice desperately needs. Think about it: how many low-value cases are abandoned because litigating costs more than the amount in dispute? How many citizens lose faith in justice because it takes years to arrive?

In my LinkedIn survey, 42% of respondents said they would prefer to have an AI resolve their $50,000 case within two weeks rather than wait years in the traditional system. That is not a love of technology; it is desperation for accessible and timely justice.

As the magistrate noted regarding consumer rights: "Low-value matters are not litigated and even less so do they reach the courts. They are left without justice." AI could be the answer to this justice gap that affects millions.

The clock is already running

"What are we going to teach at the National Judicial Training School? It is no longer enough to offer the course on constitutionality," the magistrate asks. The new judicial administration body, which will assume its functions on September 15, faces a crucial decision: does it embrace this transformation or let it die through inaction?

"We are already behind," Juan Jaime insists. And he is right. While we debate whether or not to use AI, other countries are already implementing digital courts, automated resolution of minor disputes, and predictive analysis of rulings.

The crack that becomes a path

This initial chisel strike in the ice of our justice system is not the end of anything; it is the beginning of everything. Jurimetrics is only the spearhead — the irrefutable argument that opens the door. Because if we accept that a machine can calculate better than a human with a calculator, what other mechanical and repetitive processes are we carrying out poorly out of sheer traditional stubbornness?

Magistrate Juan Jaime did not merely issue a ruling; he planted a seed of necessary transformation. The question now is not whether Mexican justice will adopt AI — that train has already left the station — but whether we will do it well, with ethics, transparency, and above all, with the courage to admit that the current system, as it stands, is no longer sufficient.

The crack has been made.

The ice is beginning to creak. And all of us who believe in a more accessible, faster, and fairer justice system should be pushing in the same direction.