Artificial Intelligence & Robotics

Judged by an Algorithm: Are judges, juries next to be totally automated?

digital gavel

Fewer than 10% of general courts are using generative AI tech or are planning on using it within the next year, according to a 2024 survey from the Thomson Reuters Institute. But 70% of legal professionals say advancements in AI and generative AI will have a transformative or high impact on the legal profession within five years. (Image from Shutterstock)

The defendant stands before the bench, heart pounding. He faces the judge and jury—only in this case, they have been replaced by a computer screen. In seconds, the artificial intelligence judge has analyzed thousands of similar cases, weighed the evidence and has come to a ruling and possible sentence. Is this a far-fetched dystopian scenario or a glimpse into the future of the legal system?

Fewer than 10% of general courts are using generative AI tech or are planning on using it within the next year, according to a 2024 survey from the Thomson Reuters Institute. But 70% of legal professionals say advancements in AI and generative AI will have a transformative or high impact on the legal profession within five years. Could that transformation extend to possibly replacing judges or jurors?

AI is already being used by courts across the country to assess risk and set bail. In 2016, the Wisconsin Supreme Court ruled in State v. Loomis that using an algorithm to assess risk at sentencing did not violate a defendant’s due process rights. The judge had used COMPAS, a risk assessment tool, to determine that defendant Eric Loomis should be sentenced to six years in prison for a drive-by shooting. COMPAS estimated Loomis’ rate of recidivism based on an interview and criminal history. In 2017, the case was appealed to the Supreme Court, but the court denied the appeal.

According to the Bail Project, more than 60 jurisdictions in the United States use a form of pretrial risk assessment tool that’s usually AI-powered.

Meanwhile, in China, AI could take on even bigger role. In December 2024, the country unveiled an AI-generated judicial platform that understands legal terms and logical reasoning.

“To put it simply, it will be a legal assistant of judges, serving them in classifying legal information and reducing their burden in handling a rising number of cases,” Zhang Chengbing, an official from the People’s Court Press, said in a statement.

So, are judges and juries the next to be totally automated?

Order in the court

Judges are already routinely using AI to draft routine administrative orders, to summarize documents, to make timelines of events and to draft opinions, says Senior Judge Herbert Dixon Jr., retired from the Superior Court of the District of Columbia and the immediate past chair of the ABA Journal’s Board of Editors.

In fact, Dixon, along with other judges and a computer science professor, created guidelines in February for navigating AI in the judiciary.

“Judicial officers and those with whom they work must ensure that any use of AI strengthens rather than compromises the independence, integrity and impartiality of the judiciary,” the guidelines states. “The use of AI and GEN AI tools must enhance, not diminish, this essential obligation.”

While these are simply guidelines that could clear the way for AI to replace judges, Dixon doesn’t believe it will happen anytime in the near future. The biggest issue at the moment, he says, is public acceptance.

He predicts that courts will take small steps, using AI to help with mediations and some arbitrations to prove it works (it’s already being used in Online Dispute Resolution, otherwise known as ODR, following a 2021 court initiative).

Gradually—he predicts that this won’t happen during his generation— using AI as a judge will gain acceptance. It will most likely begin with brief, document-oriented proceedings. This would eliminate issues related to credibility, Dixon says. He also predicts it could be utilized in routine hearings, where the decisions are based on checking a box to decide whether the statutory provisions apply or not.

The advantages of using AI as a judge include efficiency, time savings and cost reduction. But there are also many significant disadvantages, including public resistance and AI bias.

Key to moving forward would be having reliable, credible testing mechanisms comparing how AI and human judges deal with the same fact patterns, says Eugene Volokh, senior fellow at Stanford, and professor of law at UCLA.

Judges exercise discretion, taking into account social, moral, policy and political considerations, while also writing their opinions in a way that satisfies the legal community, which is not what one would expect from AI judges, explains Eric Posner, professor of law at the University of Chicago.

But if it turns out, in blind-reviewed competitions judged by respected evaluators, that some AI software routinely yields results that are at least as good or better than human judges, then many people will shift from the much more expensive human judging to the less expensive AI judging, Volokh says.

AI would lose because when it comes to credibility, empathy and mercy, it fails, says Judge Roy Ferguson, a special/private judge for family and civil cases. He says that AI is useful for polishing drafts, creating transitions between topics, reformatting existing writing into different formats and turning a brief into an order, among other tasks.

But recently, Ferguson literally asked AI if it posed a risk to the justice system, and he was chilled by the response.

“It strongly said that it did, and it gave me 10 ways that it could occur,” Ferguson says. “It came up with ways that AI could manipulate judges, witnesses, juries, law enforcement and communities,” he says.

And when Ferguson asked whether AI should be relied upon by anyone in the justice system, it gave a one-word answer: “no.”

Algorithm duty

It’s no surprise that a randomly assorted group of a dozen people might not be the best people to judge a trial. A study by Shari Seidman Diamond, a psychologist and lawyer who holds a joint appointment at the American Bar Foundation and Northwestern University, finds that in 17% of the cases, jurors didn’t understand instructions.

But would an AI jury be any better? Law firms are already using AI to predict the outcome of jury trials, and expert witnesses are using AI to recreate auto accident demonstrations for the jury, says Marc Matheny, an attorney in Indianapolis. Predictive AI can be used to generate trial strategies, the selection of juries and opening statements.

That’s why Eugene Stratton Horres Jr., a retired civil trial attorney who is a national speaker on the topic of AI, believes that jury trials will continue to be an amalgamation of human judgment and AI assistance.

While AI replacing juries in the future is possible, it would be highly controversial and fraught with challenges, he says.

AI will be able to reason, but its reasoning capabilities are different from human reasoning, which could make it difficult to serve as a jury.

“AI’s reasoning ability is based on patterns, algorithms and data, while human reasoning involves complex cognitive processes, emotions, intuition and experiences,” Horres says. AI also lacks morality, human judgment and community standards. So while AI will streamline and make some legal processes more efficient, removing and replacing human jurors altogether would risk eliminating the empathetic and moral considerations that are the cornerstones of the modern judicial system, Horres says.