Judges at Everlaw Summit '25 Say AI Cannot Replace the Human Judge

A judiciary that moves slowly on technology

The panel was held on the final day of Everlaw Summit '25 and moderated by Everlaw CLO Gloria Lee. It featured Judge Joy Flowers Conti (W.D. Pa.), Judge David Cunningham (Superior Court, Los Angeles County), Judge Willie Epps (W.D. Mo.), Judge Young Kim (N.D. Ill.) and Judge Victoria Kolakowski (Superior Court, Alameda County).

The discussion began with the gap between the pace of legal technology and the caution of the courts. Judge Conti, described by Everlaw as a veteran of the ediscovery world, said: "I'd say the courts historically have lagged about ten years behind what is happening technology-wise in the legal profession."

Judge Epps recalled that when he joined the bench in 2017 he went to court with three ring binders full of paper, until the administrative office bought OneNote and the court went paperless. Judge Kolakowski pointed to data security as a factor slowing adoption. She said people worry about outside intrusions and about the idea that information put into a generative AI system becomes public, which she said keeps judges and court staff from using powerful new tools for fear of compromising confidential case information.

AI as a tool for overworked courts

Despite that caution, the judges saw real promise in generative AI, particularly for high-volume and complex cases. The sentiment, according to Everlaw's account, was that AI is not a replacement but a force multiplier.

Judge Cunningham said filings in Los Angeles County have doubled since the pandemic and that many judges are overworked. He cited mass tort cases, many of them tied to wildfire litigation, and said AI could be useful in organizing that volume of data. Judge Conti added: "Justice delayed is justice denied, and if AI can help reduce the timeline for some of these cases, it can be extraordinarily helpful."

Judge Kim described his own use of AI tools: summarizing long opinions and documents, creating timelines, and drafting small paragraphs on standards he is familiar with, so he can easily verify the output.

The judges also stressed trust and transparency. Judge Conti said she worries about people no longer trusting the judiciary, and that transparency is very important in the early stages of generative AI adoption. Judge Cunningham, who has been appointed to chair an American Bar Association working group, called for education on how AI affects ethical obligations, particularly competence and attorney-client privilege. He said that in California, "we're trying to take the lead on this."

Deepfakes and the court's fact-finding role

The panel's most sobering segment concerned deepfakes. Judge Kolakowski described a recent case in which a self-represented litigant submitted clearly fabricated videos. One was a Ring camera video in which the background was black and white while the people in front were in color. In another, someone seemed to talk endlessly.

The concern, according to the Everlaw summary, is that generative AI is improving quickly and may make it hard for courts to tell what is real, especially without forensic resources. Judge Kolakowski said: "The hard part is that we tend to believe what the computer tells us. We tend to believe what our eyes show us."

Why the human element stays central

Closing the panel, the judges unanimously affirmed, per Everlaw, that no technology can replace judicial discretion, judgment and empathy. Judge Kim said efficiency matters, but judges must also be concerned about perception: "for the third branch to have credibility, we have to be perceived as fair."

Judge Epps said generative AI could democratize access to justice by improving the quality of filings from people who cannot afford representation. He also cautioned: "There's a lot of discretion that we have to exercise as judges, and the human element is indispensable."

Judge Kolakowski drew a line between questions of law and questions of fact. She said she cannot imagine a machine deciding questions of fact in the near future, and that a human, whether a jury or a judge, needs to evaluate the evidence.

For legal professionals, the message from the bench is that AI use is welcome where it saves time and can be verified, but that transparency, education on ethical duties and human oversight of evidence remain essential.