A New York-based lawyer has been fined after he misused the artificial intelligence chatbot, ChatGPT, relying on it for research for a personal injury case.
Last week Steven A. Schwartz, fellow attorney Peter LoDuca and law firm Levidow, Levidow & Oberman, were fined US$5,000 (AU$7,485) for submitting fake citations in a court filing.
The judge found the lawyers acted in bad faith and made “acts of conscious avoidance and false and misleading statements to the court.”
In a written opinion, Judge P. Kevin Castel said attorneys had to ensure their filings were accurate, even though there was nothing “inherently improper” about using artificial intelligence in assisting with legal work.
“Technological advances are commonplace and there is nothing inherently inappropriate about using a reliable artificial intelligence tool for assistance,” Castel wrote.
“But existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings.”
Schwartz, who has more than 30 years’ experience practicing law in the US, was part of a legal team acting for a man suing the airline Avianca. The client, Roberto Mata, had claimed that he was injured after a metal serving cart hit his knee during a flight.
Unfortunately for the client, Schwartz did his legal research for the case using ChatGPT without fact checking if the cases he cited in his brief, involving other airlines and personal injuries, were real or not.
Turns out they weren’t.
“He did ask ChatGPT whether one of the cases was real but was happy enough when ChatGPT said yes,” Professor Lyria Bennett Moses tells ABC RN’s Law Report.
“In fact, ChatGPT told him that they could all be found on reputable databases, and he didn’t do any checking outside of the ChatGPT conversation to confirm the cases were real — he didn’t look any of them up on a legal database .”
Professor Moses is the director of the UNSW Allens innovation hub. She says the lesson here is to use this platform with caution.
“[Schwartz] stated in the court [hearing]’I just could never imagine that ChatGPT would fabricate cases.’ So, what it showed was a real misunderstanding of the technology,” he explained.
“[ChatGPT] has no truth filter at all. It’s not a search engine. It’s a text generator working on a probabilistic model. So, as another lawyer at the firm pointed out, it was a case of ignorance and carelessness, rather than bad faith.”
Schwartz’s lack of due diligence when researching his brief in this personal injury case has caused him great embarrassment, particularly as his hearing has drawn worldwide attention.
After reading a few lines of one of the fake cases aloud, Judge P. Kevin Castel asked: “Can we agree that’s legal gibberish?”
When ordering the lawyers and the law firm to pay the fine, the judge said that they had “abandoned their responsibilities when they submitted non-existent judicial opinions with fake quotes and citations created by the artificial intelligence tool ChatGPT, then continued to stand by the fake opinions after judicial orders called their existence into question.”
In a statement, the law firm Levidow, Levidow & Oberman said its attorneys “respectfully” disagreed with the court that they had acted in bad faith.
“We made a good-faith mistake in failing to believe that a piece of technology could make up cases out of whole cloth,” it said.
Lawyers for Schwartz told Reuters that he declined to comment while lawyers for DoLuca were reviewing the decision.
Varying opinions on AI
Schwartz’s mishandling of artificial intelligence has rippled across the US and the world.
For example, a judge in Texas has now insisted that any attorney appearing in his court must attest that no portion of a filing was drafted by generative artificial intelligence or, if it was, that the filing had been fact checked by a human being.
However, not all judges are opposed to using chatbots in the courtroom.
Judge Juan Manuel Padilla, who is based in Colombia, revealed in a judgment that he had consulted ChatGPT in a case involving an autistic child.
“He was concerned about whether an autistic child’s insurance should cover all the costs of his medical treatment,” Professor Moses says.
“For this case, he went and asked ChatGPT if an autistic minor was exempt from paying fees for their therapies and ChatGPT replied, ‘yes, this is correct’.”
According to the regulations in Colombia, minors diagnosed with autism are exempt from paying fees for their therapies.
“Now, in that case, the judge didn’t simply copy that ChatGPT text and call that his judgment. He used other precedents to support his ruling,” he explains.
“And he continues to advocate for the usefulness of this tool for judges in order to enhance efficiency in court processes and judicial decision making.”
More to be learned
Professor Moses believes there is much to be gained from using AI chatbot technology, but there needs to be an understanding of what its limitations are.
“The mistake Steven Schwartz made was just a misunderstanding of what ChatGPT was — he was thinking of it a little bit like a sort of enhanced Google search engine,” he explains.
Additionally, Schwartz didn’t reveal that he used ChatGPT in his submission, Professor Moses says, which he should have. It’s also important that employees aren’t charging clients for work they’re not doing themselves or fact checking — citation and transparency are key.
“You’ve also got to be careful with the information you input into ChatGPT so don’t put confidential client information into your prompt,” she says.
Be aware of the risks and the limitations, she adds. But that’s not to say that it can’t be a useful tool in some contexts, and it might become more useful over time.
RN in your inbox
Get more stories that go beyond the news cycle with our weekly newsletter.
