AI-powered robot lawyer heads to court in first test to disrupt the legal system
Kaelan Deese
Video Embed
The so-called “world’s first robot lawyer” powered by artificial intelligence will soon help a defendant fight a traffic ticket in court next month, though its creator aspires to one day bring it to the halls of the highest court in the land.
That’s according to Joshua Browder, CEO of DoNotPay, who told the Washington Examiner his AI attorney will soon participate in two hearings, one in-person and one over a virtual Zoom call. The company’s AI runs on a smartphone, listens to court arguments, and forms responses for the defendant.
MICROSOFT EYES $10B INVESTMENT IN CHATGPT DEVELOPER
“For the in-person hearing, the AI is going to be whispering in someone’s ear with AirPods what to say. And for the online hearing, we’re thinking about a teleprompter, but we are also considering a synthetic voice. So it’s all up in the air at the moment,” Browder said.
Although DoNotPay has not yet tested its technology in the courtroom, it has already used AI-generated form letters to help people gain refunds for disputed bills and parking tickets. The company claims to have succeeded in helping over 2 million customers with service disputes and court cases between other organizations and institutions.
The idea behind DoNotPay was inspired in part by statistics from the American Bar Association, which estimates as much as 80% of the legal needs of low-income citizens go unmet. The company has raised nearly $28 million from tech-based venture capital firms such as Andreessen Horowitz and Crew Capital.
“The legal system right now is pay-to-play. And so we’re just making sure everyone is equal. It’s up to the judge to decide either way, how harsh or whether they’re guilty or not, or maybe a jury one day, if the case does get more sophisticated, but everyone at least deserves a defense,” Browder said.
While Browder wants to shake up the judicial system, court watchers have cast severe doubt on the practicality and morality of robots replacing lawyers.
Nicholas Saady, a litigator for Pryor Cashman who offers insight on using AI in business and in the legal world, said that DoNotPlay’s plan is at risk of violating state laws requiring lawyers to be licensed.”Is it the unauthorized practice of law?”
Saady presented hypothetical examples of why using AI in a court of law could pose a detriment to a defendant, such as an inability to read body language. “It doesn’t seem like AI is ready to get on its feet in court,” Saady told Politico.
When asked whether he thinks real-world attorneys might be uncomfortable with the prospects of AI doing their work, Browder said he believes “a lot of lawyers are safe.”
“I’m friends with some human rights lawyers and people doing very serious work arguing important cases and I think they don’t have to be worried,” Browder said. “But I think a lot of the billboard lawyers will get replaced. If you’re in a car accident and you need to sue someone for $10,000, then you can definitely do that with AI.”
DoNotPay has vowed to cover the cost of any fines if its AI loses the case. The company uses GPT-J, an open-source AI model released last year, as well as models from OpenAI, which is working on some of the state-of-the-art AI tools such as ChatGPT and could soon attract investments valuing it at $29 billion, according to the Wall Street Journal.
“Unfortunately, the AI lies a lot … we’ve had to make sure that the AI doesn’t lie, so in our prompts to it, we say, ‘stick to the information provided’ and things like that,” Browder said.
Browder declined to reveal the name of the client and the court where the hearing will be next month, citing online “threats” he’s received since unveiling his plan. “We’re too scared to reveal it,” he said, adding his technology is “in the letter of the law but not in the spirit of the law or the court rules.”
Widespread use faces significant barriers since the technology isn’t legal in most courtrooms. There are states that require all parties to consent to being recorded, which limits the ability of robot lawyers to enter many courts of law. For example, of the 300 cases Browder’s company has considered for a trial of its AI, only two were permitted.
“I think with these two cases, it will be a very powerful proof of concept [and] my goal is in the next 24 months to have five states explicitly allow AI in the courtroom to help people,” Browder said, noting states such as Utah and California have established “task force” panels to better understand whether AI can be utilized in their respective state courts.
Browder sustained an onslaught of criticism from legal minds on Twitter this week when he proposed a $1 million offer for any attorney willing to allow his AI to argue a case pending before the Supreme Court.
Several critics noted technology such as AirPods and phones are not allowed in the high court when justices argue cases and that Browder’s monetary offer would need to be much larger to attract the interest of a star law firm with a case before the court due to the sheer rarity of getting the justices to take up a case.
But Browder told the Washington Examiner that just after his viral tweet on Monday, he had received offers from lawyers in cases from “appeals courts, district courts, and state courts.”
CLICK HERE TO READ MORE FROM THE WASHINGTON EXAMINER
Another major complication surrounding the novel technology of OpenAI is its consistency on factual matters, as it will occasionally make errors. Browder said his AI has been retrained based on seven years of data to become an “expert at the law,” noting its education is probably “like a mid-level college graduate.”
“So now we need to send it to law school,” Browder said.