My lawyer, the robot

With help from Derek Robertson

Call it the Bergerac’s Cyber ​​​​Year Protection.

The new power of artificial intelligence will be shown in the courtroom – in the form of an AI chatbot lawyer who will soon argue a case in the traffic court.

According to Joshua Browder, the founder of the consumer empowerment startup that started the program.

Sometime next month, Browder plans to send a real defendant into a real courtroom armed with a video recorder and a set of headphones. Browder’s company will feed the audio of the proceedings into an AI that will listen to legal arguments; the defendant, he said, agreed to repeat the words of the chatbot to an unknown judge.

Browder refuses to identify the defendant or the judge for the court date next month, because he is worried that the judge might get wind of the planned action and block it.

In recent months, the public announcement of advanced AI tools has raised questions about everything from higher education to the use of what should be human. Now the technology is ready to meet legal and public policies that do not expect a role for non-humans – especially in court.

All areas of technology are “cross-cutting,” Browder said. where AI sows tools and laws to determine the balance of power between individuals, on the one hand, and governments and organizations, on the other. He sees AI as a way for ordinary citizens to level the playing field in the fight against fines and fees, which they often cannot afford to judge themselves.

AI has already been introduced to the American legal profession, which is often used by large companies to help with the task of reviewing documents that can number in the millions at the time of obtaining a judgment.

But Nicholas Saady, a litigator at Pryor Cashman who advises on the use of AI in business and law, says this latest proposal may not fly. He pointed to many examples and specific cases presented by using AIchatbots for real time as legal representatives. he asked, saying the plan risks violating state laws requiring professional licenses for attorneys.

Browder said he has reviewed this and informed two authorities that his plan “is not legal.” In the second administration, he said, he arranged for a defendant to use AI in the Zoom trial.

For the Zoom hearing, Browder said, instead of asking the accused to speak out loud themselves, he is considering doing the trial in one step by using an AI tool to able to imitate a person’s voice after recording their speech. “Although,” he admitted, “it can get us into a lot of trouble.”

This is not Browder’s first foray into solving technology disputes. Browder’s company offers all kinds of DIY legal and consumer assistance, and he has been using AIs to help consumers spread their claims to governments and organizations since 2021, mostly in the creation of books and data for online discussions. One of the problems he faced along the way was that AI would sometimes do things to push their case, he said. To prevent this, he said, his company should take special measures to encourage bots to stick to real information.

For his courtroom, Browder says he’s using GPT-J, an AI model introduced last year. For advanced applications, Browder said his company uses models from OpenAI, the company behind the latest AI tools, called big language models.

Browder said he remains in contact with OpenAI to ensure that his company, DoNotPay, does not violate its terms of use. In practice, this means adding features – such as a two-second delay in when a user can reject an AI-suggested answer – to ensure that the user maintains the order. the final results, he said.

OpenAI did not respond to requests for comment.

Browder’s courtroom gambit is one of the most recent curveballs AI-powered tools have thrown at the legal profession. A peer review pre-printed paper published last week by two law professors predicting that a multilingual sample will soon be able to pass the multiple-choice section of the Multistate Bar Exam.

But even if an AI can pass the Bar, and even if the courts allow AI jurors, Saady says it would be better to hire flesh-and-blood lawyers.

Good judgment, he said, depends on things that cannot be seen by bots: He cites the ability to read body language, and making split-second trial decisions in the middle of courtroom deals. “AI doesn’t seem ready to stand on its own two feet in court,” he said.

Boston attorney Matt Henshon, on the other hand, says the concept of AI-powered attorneys holds promise. Henshon, the chairman of the American Bar Association’s Artificial Intelligence and Robotics Committee, said that legal aid can be obtained in low cases where a person may go without a representative.

In that context, he compared AI attorneys to class-action lawsuits, where many plaintiffs can obtain precedents by combining many small claims that are not valid on their own.

“There are many legal mistakes that cannot be corrected because it is useless to involve a lawyer,” said Henshon.

Browder — whose father, Bill Browder, spearheaded the passage of the Global Magnitsky Act and other laws around the world that sanctioned human rights abuses — said he hopes his techno-legal use of the a help to make that case in the state of the state and other organizations that make laws. they will take advantage of the rise of AI lawyers.

“There are all these gatekeepers, there are laws, there are governments that work together,” he said. “It’s not the technology that’s the hardest. It’s the compliance.”

Stablecoins may be very close to the traditional banking system for the comfort of the latter.

As Sam Sutton of POLITICO noted at this morning’s money newsletter, the Banking Policy Institute, which is led by JPMorganChase CEOJamie Dimon, is raising concerns about a plan that would allow the leading stablecoin issuer Circle to lend its reserves to the Fed in exchange for Treasury bonds securities and interest.

“” The implications here are hard to overstate. This means that anyone in the world who wants to hold equity in the Federal Reserve will be able to do so by buying (Circle’s stablecoin) USDC. Especially in times of crisis, this asset will be most needed as a back-end CBDC or an account at the Federal Reserve,” wrote BPI staff in their post.

As Sam said, regulators have long been concerned about subjecting stablecoins and other crypto assets to regulations before attaching themselves to traditional currencies. — Derek Robertson

What if the next thing is to generate text AIs Could something a bit more of a refresh be in order?

In a blog post this morning, George Mason economist and prolific blogger Tyler Cowen posits that ChatGPT could function much in the same way as one particular innovation of the George W. Bush-era web: The blog comment.

“Every time I ask a question (in ChatGPT) it’s like putting a blog post out there… and getting a barrage of answers right away, and without having to post it.” (his blog) and a lot of my thoughts every day,” Cowen wrote. “…I learn from the answers given, and what’s more, some of them are very good and accurate. good too.”

The rating reflects Cowen’s deep understanding of what a product like ChatGPT actually does – that is, it uses all the information it has on the website, and return it in a way that mimics how it was written by a human. Examples like ChatGPT “are like giving everyone their own information-open blog, with examples of super-infovores as readers and quick responses and follow-ups when needed,” he added. eat (Note that he said “follow-up,” however, he didn’t communicationthe essential human-to-human element of the blog that made Twitter exist.) – Derek Robertson