California courts hope to harness artificial intelligence

SAN FRANCISCO (CN) — In the emerging technology of artificial intelligence, getting your head around it is akin to wrestling with an octopus.

That’s how Mary J. Greenwood described it at Friday’s Judicial Council of California meeting. A member of the council’s AI Task Force, Greenwood discussed creating policy for using generative artificial intelligence for the state court system.

One job of her task force is to identify specific issues concerning AI.

“It’s a little like wrestling with an octopus in addition to being covered in slime,” Greenwood said to laughs.

After what she called scared looks from task force members, understanding began to creep in as they learned more about the technology.

The group — which hopes to present the policy in July for council approval — identified four focus areas, the first of which is a model use policy and guidance for courts.

This is key, Greenwood said, as a survey issued to courts about AI revealed that many of them are waiting on the council to provide them guidance.

“Everyone has the same concerns on their mind,” said Greenwood, the administrative presiding justice of the Sixth District Court of Appeal.

AI should help and promote the administration of justice but also have safeguards for privacy while maintaining public trust, she added.

Part of that includes information court users might give AI.

Information given is used to train an AI and could potentially be provided to another user unless there are guardrails in place protecting private details. Greenwood said she doesn’t want critical information released into what she called “the vortex.”

“Not ‘The Matrix,’” she added, referring to the move in which AI is a controlling, ubiquitous enemy of mankind. “The vortex.”

AI also must avoid hallucinating, or creating false content and presenting it as accurate. Greenwood referenced an attorney who cited a nonexistent law provided by AI.

That’s happened more than once. In one case, a judge noted that six cases listed in a legal filing appeared fictitious and included false quotations and citations.

“Hallucinations include completely inventing case names,” Greenwood said.

The policy also will target the removal of bias from AI.

According to Greenwood, an AI was asked to provide a picture of a group of research attorneys. The picture offered contained only white men, a representation Greenwood said doesn’t accurately portray the court system.

The other three aspects the task force will address include AI as a self-help tool for court users, ensuring AI isn’t misused to affect court evidence, and determining how the tool could impact legal research.

After a monthlong public comment period starting in March, a successful July council vote on the policy would make it effective Sept. 1.

“We are obviously far from being done,” Greenwood said, adding moments later: “There is still a great deal to be done.”

AI is a topic that’s also drawn the interest of California’s other two branches of government.

Governor Gavin Newsom in September 2023 issued an executive order about AI. It focused on how the state will use the technology while protecting itself and Californians.

In response, a year later, state workers invited AI industry representatives to a showcase where they delivered pitches. One state official said he hoped AI could help comb through lengthy reports, finding the essential information from a long narrative and saving someone’s time.

Lawmakers also see AI as an area that needs regulation.

Bills introduced this session on the issue run the gamut. State Senator Scott Wiener, a San Francisco Democrat, has a bill that would create safeguards for developing AI frontier models — those on the cutting edge.

State Senator Jesse Arreguín, a Berkeley Democrat, introduced legislation that would require official reports written by law enforcement with AI’s help to include a disclosure that the technology was used.