Artificial Intelligence, Meet the U.S. Supreme Court.
This is an organization steeped in tradition and resistant to rapid changes in the way things are done. But whether they like it or not, the judges are about to watch an artificially created version of themselves, essentially an avatar, speak the words they actually spoke in court, but could not be heard at the same time by anyone other than those in the courtroom.
Northwestern University professor Jerry Goldman has been devising ways to make the Supreme Court more accessible to the public since his nonprofit project Oyez went live on the Internet in 1996. The site dates back to 1955, when the Supreme Court began recording its proceedings, and aimed to provide audio of the court’s oral arguments and opinions in every case it decided.
Professor Goldman’s Oyez project received a lot of attention when it debuted, as the public did not know until the early 1990s that courts were recording courtroom proceedings. And because the tape preservation process was so complicated, many of those recordings were lost forever. Furthermore, access to audio was severely restricted. In fact, no one outside the courtroom was allowed access until several months after the case was heard and decided. It was not until the beginning of the next court period that the audiotapes of the previous period became generally available.
In 2020, when the country was shut down due to the coronavirus pandemic, courts were essentially forced to allow all oral arguments to be broadcast live, with the justices connected by telephone lines and the public able to listen in. And in the wake of the pandemic, justices who had long resisted regular audio broadcasts of their arguments, left the pandemic system in place.
As a result, only some of the court’s official functions are no longer available on the same day. The announcement of a highly important decision is summarized by a judge from the court and, in some cases, includes an oral dissent. To this day, the old system of restricting access until the next term remains in place, with only those actually in the courtroom able to hear or see the day’s drama.
Now, Goldman’s team is experimenting with making the drama even more realistic, even though audio hasn’t been available for months. They use AI to recreate what people in the courtroom saw, not just heard, when the verdict was announced.
As Goldman says, “If it’s open to the public in court, it should be open to everyone. That’s simple.”
But since cameras aren’t allowed in the Supreme Court, how does Goldman’s new site, On The Docket, create its visuals?
Answer: Basically, we use avatars. And it wasn’t easy, says University of Minnesota professor Timothy R. Johnson, one of the project’s architects along with AI design firm Spooler. He says some of the early AI efforts were interesting.
“Some failures have shown that when you give the robot certain commands, it does something downright creepy, like a particular judge magically disappearing from the bench or all the judges bending over at the same time.”
Ultimately, by using photos and videos of judges in public, we were able to create realistic-looking video versions of each judge, including their demeanor, head tilt, and hand gestures. They were able to match these avatars with existing real audio.
Of course, there were also ethical issues they had to grapple with. Should the video look completely real, or should we do something special to let the viewer know it’s not? In the end, they decided to slightly cartoonize the video and clearly mark what was generated by the AI so viewers could tell what was real (audio) and what wasn’t (video).
For example, in their first foray into this brave new world, they created a visual for Chief Justice John Roberts’ 14-minute summary from the court’s 6-3 decision granting then-former President Trump and all future former presidents complete immunity from prosecution for any core official act committed while in office, no matter how egregious.
Roberts was followed by Justice Sonia Sotomayor, who outlined a dissenting opinion on the court.
Their passionate storytelling, which lasts a total of 38 minutes, is both captivating and a little creepy.
The courts probably won’t like this latest depiction of reality. Ultimately, the recordings of the oral arguments and presentations were kept secret until 1993, when law professor Peter Irons signed an oath not to release the then-secret recordings, and then published the oral arguments in a book containing a dubbed cassette of the arguments, which he considered so important.
The court immediately sued him, but it appears that the suit was dismissed as soon as it was determined that the case had been defeated.
Since then, thanks to the coronavirus pandemic, oral arguments have been broadcast regularly, but the court’s announcement of the court’s interesting decision was kept secret until months after it was handed down. Reporters have repeatedly requested that their opinion presentations be broadcast live, and academics have also made similar requests.
For example, Professor Goldman points out that he reviewed Warren Court documents from the early 1950s in which the justices discussed memorializing oral arguments and opinions with tape recordings.
“There’s no indication in these papers that they want to keep these secret,” he says.
But as far as is known, all requests from reporters and scholars that courts make court presentations available as live audio broadcasts have been met with deafening silence. And even AI can’t capture it as audio or video.
Copyright 2026 NPR
