Chatting with the Ninth Circuit judicial conference last week, Justice Elena Kagan complimented Anthropic’s proprietary AI bot Claude for its analysis of the Confrontation Clause.
This might be the strongest policy argument for court expansion yet. We desperately need to get this woman some more co-workers. Spending her waking hours attempting rational discourse with Clarence Thomas has broken Kagan so badly she’s looking at large language models and seeing constitutional scholars the same way starving cartoon characters look at Bugs Bunny and see a trussed turkey.
Kagan’s remarks were inspired by an experiment conducted by Jenner & Block’s Adam Unikowsky, employing Claude 3.5 Sonnet to perform a number of analytical tasks following Smith v. Arizona. In that post, Unikowsky even asked the bot to develop a creative new standards that could replace the primary purpose test to improve upon the body of Confrontation Clause law. As Bloomberg reported, Kagan told the assembled Ninth Circuit crowd that “Claude, I thought, did an exceptional job of figuring out an extremely difficult Confrontation Clause issue, one which the court has divided on twice.”
More recently, Unikowsky set up Claude to conduct a mock Supreme Court oral argument based on one of his actual oral arguments. Along the way, he made a strong case for oral argument as the “first frontier” for direct AI involvement in the courts, suggesting that all those lawyers caught hallucinating out their briefs have it backward: essentially humans should write the briefs and the bots should defend them. It certainly gives the bot more expert guidance, though it still seems like an idea that’s all well and good until a justice invites it to start talking about white genocide.
That’s only slightly sarcasm. Remember when Sam Alito asked a series of questions based on the batshit premise that because the statute making certain abortions legal used the phrase “unborn child,” later abortion bans using that same wording should render the first statute null. Or something. But even though in Unikowsky’s experiment the algorithm held its ground against a dumb question before trying to chart a reasonable path between the Scylla & Charybdis of a bad faith judge — we can call it the Scalia & Clarencybdis effect — it’s easy to see how a judge could use flawed premises or invented facts to trick a bot into damaging answers.
There are, of course, mechanisms to protect against this… on the other hand, they just found out that a string of three-digit numbers can subliminally convince generative AI to become a homicidal owl-lover, so the guardrails may be more paper thin than we think.
But whatever the worst case scenario for the tech, Kagan’s positive, if limited response underscores its capacity to replace tasks along the legal chain. Career coach Jane Genova compares it to LegalZoom:
The implications for employment of all lawyers should alarm. Recall how online service LegalZoom wiped out myriad types of Main Street lawyers who handled routine legal matters for individuals. Later, it expanded its services to small businesses. Will SCOTUS Justices be hiring more AI robots and fewer human clerks?
Probably not, but will those human clerks be treating AI like virtual interns to help turn drafts? Probably so. And probably soon. Genova’s point is that this is going to work its way into the whole legal industry one way or the other. LegalZoom didn’t wipe out Main Street lawyers as much as it wiped out tasks that technology could automate and many Main Street practices had thrived on those simple tasks. Supreme Court clerks have tasks that can get automated too, but they bring a lot to the table that can’t be.
Everyone’s talking about hallucinations right now, but once users understand how to reliably prevent this technology from injecting its own drunken bullshit, it’s actually a decent tool. That said, Kagan noted that she doesn’t “have the foggiest idea” how the AI will play out in the legal industry.
Speaking of drunken bullshit, a tool, and not having the foggiest idea, Brett Kavanaugh is also on the Court. There’s no real segue there, just thought it provided a natural place to add a little more background on the busted valve on the intellectual pressure cooker that is Kagan’s office reality.
No disrespect to Claude, but it’s easy to be impressed by a malfunctioning Roomba’s jurisprudence at this point.
A brief history of the Confrontation Clause [Adam’s Legal Newsletter]
Automating oral argument [Adam’s Legal Newsletter]
Earlier: You Can Replace Supreme Court Lawyers With AI Now. Honestly, That Tracks.
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news. Joe also serves as a Managing Director at RPN Executive Search.
The post Elena Kagan Praises Artificial Intelligence Now That She Works With So Little Human Intelligence appeared first on Above the Law.
Chatting with the Ninth Circuit judicial conference last week, Justice Elena Kagan complimented Anthropic’s proprietary AI bot Claude for its analysis of the Confrontation Clause.
This might be the strongest policy argument for court expansion yet. We desperately need to get this woman some more co-workers. Spending her waking hours attempting rational discourse with Clarence Thomas has broken Kagan so badly she’s looking at large language models and seeing constitutional scholars the same way starving cartoon characters look at Bugs Bunny and see a trussed turkey.
Kagan’s remarks were inspired by an experiment conducted by Jenner & Block’s Adam Unikowsky, employing Claude 3.5 Sonnet to perform a number of analytical tasks following Smith v. Arizona. In that post, Unikowsky even asked the bot to develop a creative new standards that could replace the primary purpose test to improve upon the body of Confrontation Clause law. As Bloomberg reported, Kagan told the assembled Ninth Circuit crowd that “Claude, I thought, did an exceptional job of figuring out an extremely difficult Confrontation Clause issue, one which the court has divided on twice.”
More recently, Unikowsky set up Claude to conduct a mock Supreme Court oral argument based on one of his actual oral arguments. Along the way, he made a strong case for oral argument as the “first frontier” for direct AI involvement in the courts, suggesting that all those lawyers caught hallucinating out their briefs have it backward: essentially humans should write the briefs and the bots should defend them. It certainly gives the bot more expert guidance, though it still seems like an idea that’s all well and good until a justice invites it to start talking about white genocide.
That’s only slightly sarcasm. Remember when Sam Alito asked a series of questions based on the batshit premise that because the statute making certain abortions legal used the phrase “unborn child,” later abortion bans using that same wording should render the first statute null. Or something. But even though in Unikowsky’s experiment the algorithm held its ground against a dumb question before trying to chart a reasonable path between the Scylla & Charybdis of a bad faith judge — we can call it the Scalia & Clarencybdis effect — it’s easy to see how a judge could use flawed premises or invented facts to trick a bot into damaging answers.
There are, of course, mechanisms to protect against this… on the other hand, they just found out that a string of three-digit numbers can subliminally convince generative AI to become a homicidal owl-lover, so the guardrails may be more paper thin than we think.
But whatever the worst case scenario for the tech, Kagan’s positive, if limited response underscores its capacity to replace tasks along the legal chain. Career coach Jane Genova compares it to LegalZoom:
The implications for employment of all lawyers should alarm. Recall how online service LegalZoom wiped out myriad types of Main Street lawyers who handled routine legal matters for individuals. Later, it expanded its services to small businesses. Will SCOTUS Justices be hiring more AI robots and fewer human clerks?
Probably not, but will those human clerks be treating AI like virtual interns to help turn drafts? Probably so. And probably soon. Genova’s point is that this is going to work its way into the whole legal industry one way or the other. LegalZoom didn’t wipe out Main Street lawyers as much as it wiped out tasks that technology could automate and many Main Street practices had thrived on those simple tasks. Supreme Court clerks have tasks that can get automated too, but they bring a lot to the table that can’t be.
Everyone’s talking about hallucinations right now, but once users understand how to reliably prevent this technology from injecting its own drunken bullshit, it’s actually a decent tool. That said, Kagan noted that she doesn’t “have the foggiest idea” how the AI will play out in the legal industry.
Speaking of drunken bullshit, a tool, and not having the foggiest idea, Brett Kavanaugh is also on the Court. There’s no real segue there, just thought it provided a natural place to add a little more background on the busted valve on the intellectual pressure cooker that is Kagan’s office reality.
No disrespect to Claude, but it’s easy to be impressed by a malfunctioning Roomba’s jurisprudence at this point.
A brief history of the Confrontation Clause [Adam’s Legal Newsletter]
Automating oral argument [Adam’s Legal Newsletter]
Earlier: You Can Replace Supreme Court Lawyers With AI Now. Honestly, That Tracks.
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news. Joe also serves as a Managing Director at RPN Executive Search.
