Now AI Can Defend You In Court

Published on

Now AI Can Defend You In Court; Robot Lawyers Can Avoid Unauthorized Practice Threat

AI Can Defend You In Court

WASHINGTON, D.C. (January 27, 2023) – Artificial intelligence [AI] – which has proven that it can already write good newspaper articles and speeches, pass tough exams at business schools and law schools, and even write poems and create valuable works of art – can now defend you in court, says public interest law professor John Banzhaf.

Get The Full Henry Singleton Series in PDF

Get the entire 4-part series on Henry Singleton in PDF. Save it to your desktop, read it on your tablet, or email to your colleagues

Q4 2022 hedge fund letters, conferences and more


Although the first trial run of the system in a real courtroom was delayed by threats to criminally prosecute the businessman behind it for the unauthorized practice of law, there's a very easy way which could probably be used to circumvent that problem, says Banzhaf, who wrote the first article about computers and the law, as well as obtaining the first copyrights on computer programs.

Joshua Browder, the CEO of a company called DoNotPay which has already raised $28 million, was ready to test his "robot lawyer" in a California court on February 22nd. Here's how it was supposed to work.

The driver challenging a speeding ticket would wear smart glasses that both record court proceedings and dictate responses into the defendant's ear from a small earpiece. The responses which the defendant was then supposed to voice to the judge in defending himself would come from an AI powered system.

But Browder says he received "threats from State Bar prosecutors" claiming that providing AI assistance to a client would constitute the criminal unauthorized practice of law, and "it seems likely they will put me in jail for 6 months" if he uses his app (which uses ChatGPT and DaVinci) in the courtroom.

But says Banzhaf, there's seems to be a very simple way to avoid violating the law, and of also doing so in such a way as to make the operation very inexpensive and therefore profitable.

Noted law professor Joshua Blackman has done a legal analysis of the broader issue of using computers in legal practice entitled "The Looming Ethical Issues for Legal Analytics."

He noted that people have long used computers to do legal research, to do mathematical analysis of the chances of success, to calculate a reasonable settlement amount, and for other data analytic purposes. He concluded that it might constitute the unauthorized practice of law if this information and/or services were provided directly to a lay client.

But, he also concluded, "as long as a lawyer somewhere in the pipeline independently reviews the data analytics recommendations, and blesses it, I don't see any of these as significant problems." So, says Banzhaf, one way for Browder to circumvent the problem might be to insert a lawyer between the AI system and the client in the courtroom.

Banzhaf suggests that Browder's firm could have the AI system listen to the court proceedings and then transmit its advice and suggestions, not directly to the defendant himself in traffic court, but rather (as a message in text form) to a member of the bar sitting in front of a computer monitor screen or cell phone in a room anywhere in the world.

The lawyer could look at the AI-generated message and, if he approves it, then transmit the suggestion - i.e. by pushing a "SEND" button - to the client. The client then reacts and acts exactly as he would have if the message had came directly to him from the AI device, rather than with a 2-second delay necessary to have a lawyer first review it.

This way there would be a "lawyer somewhere in the pipeline" so that the advice to the client would technically be coming from a lawyer, and not from an AI-powered computer - which only made a suggestion; one which the lawyer approved and then chose to transmit to the client.

In other words, it's analogous to situations in which lawyer relies upon upon research - including even, for example, calculations of possible settlement amounts, finding similar cases, percentage of similar cases which were won, etc. - to advise a client on what to do.

Banzhaf has also suggested to Browder how to go about initially testing the system in traffic court.

He says that, in addition to the client with his special glasses, and a lawyer at a remote location ready to read and immediately approve the AI-generated advice, there should be a live standby second-chair lawyer in the courtroom to step in and do whatever might be necessary, depending on what happens and then only to the extent necessary.

Also, initially, the lawyer at the remote location should be one very competent in offering representation in this specific field and venue, and also well practiced in reading and evaluating the AI-generated recommendations, and then very promptly pushing the "SEND" button to allow them to be transmitted to the client in court.


While having a standby attorney in court - as well as a very competent attorney at a remote location to evaluate and pass along the AI suggestions - would be very expensive, it would only be necessary during the initial phase when the system is being tested and becoming accepted.

However, once the system is up and running after being tested and generally accepted, a single lawyer could probably be able to watch several different computer monitoring screens involving several different trials, all at one time.

Then, rather than having to employ a highly trained lawyer for the remote approval process, the firm could probably hire a recent law graduate who still has no legal position, a retired lawyer seeking to make a small amount of money on the side, a new mom or other lawyer who wants to work from home only during certain hours, etc., for much less than employing a skilled practicing attorney would require.

Indeed, such an attorney could work from anywhere, since his role could be performed easily on a PC or cell phone from any location. Moreover, at this point, there would be no need for a standby attorney to sit by the client in court.

Banzhaf suggests that, so long as the AI advice is no worse than what is currently being offered by the lawyers forced to ply their trade in traffic court, the firm could be successful, and possibly go on to more difficult cases.