A Supreme Court judge ponders the impact of AI on law

US Chief Justice John Roberts A report published at the end of the year warned that the technology could “dehumanize” the law, citing the massive rise of artificial intelligence by 2023.

“Proponents of AI have the potential to increase access to justice, especially for litigants with limited resources,” they wrote. “AI clearly has great potential to dramatically increase access to key information for lawyers and non-lawyers alike.

But it clearly invades privacy interests and can dehumanize the law, he continued.

Roberts points to the increased use of AI in smartphones, voice recognition software, and smart TVs—but also points out how law professors are seeing technology emerge with admiration and caution.

Much of the analysis focused on the capacity of the legal system to make it accessible.

“For people who can't be lawyers, AI can help,” Roberts said. It powers new and highly accessible tools that answer basic questions, including where to find templates and court forms, how to fill them out and where to submit them to the judge – all without leaving home.

The Chief Justice's term comes at a time when artificial intelligence, particularly generative artificial intelligence, has penetrated many industries and used issues including education, defense, health care and the legal system.

“These tools have the welcome potential to bridge the gap between resources and urgent needs in our court system,” Roberts continued. But any use of AI requires caution and humility.

Judge Roberts pointed out that AI's limitations, including hallucinations, have led lawyers to talk about cases that don't exist.

In October, former Fugees member Pras Michel, a lawyer at the former legal team, asked for a new trial, arguing that the use of generative AI and the technology's fact-finding practice had caused his client to lose his case.

In December, a federal judge asked former President Donald Trump's lawyer, Michael Cohen, to show evidence published in court filings after the court said he could not prove their existence.

Generative AI developers have invested heavily in fighting AI nightmares. In May, OpenAI said it was improving Chat GPT's mathematical problem-solving skills to reduce illusions. In December, Fetch AI and SingularityNET announced a partnership to curb AI nightmares using decentralized technology.

“SingularityNET is working on several methods to solve the problems in LLMs,” said Alexey Potapov, SingularityNET's Chief AGI Officer. “We've focused on this since SingularityNET was founded in 2017.”

“Our view is that LLMs can only go as they are and not enough to take us to artificial general information, but they can distract from the final goal,” Potapov added.

For his part, Roberts highlighted potential biases built into AI models that could lead to unfair decisions in court cases.

“The use of AI in criminal cases to evaluate decisions involving flight risk, recidivism and other predictors has raised concerns about process, reliability and bias,” Roberts said. “At least for now, studies show a persistent view of the ‘human-AI fairness gap,' which reflects the view that human judgment, with all its flaws, is fairer than anything the machine spits out.”

Despite the caveat, Roberts remains optimistic that artificial intelligence won't replace human judges anytime soon.

“But with equal confidence, I predict that the judiciary — especially at the trial level — will be significantly affected by AI,” Roberts said. “These changes include not only how judges do their jobs, but also how they understand the role of AI in the cases before them.”

Edited by Ryan Ozawa.