We know that they are. Sometimes it jumps off the page. In fact, here's an article detailing how much it's being done:
To get a baseline, I tested all the published circuit opinions from January 2022, more than 300. Pangram found no signs of AI-generated text in any of them. Not one passage, and not one sentence. Every opinion returned "0.000000%" AI. That gave me some confidence that Pangram was not mistaking the ordinary conventions of judicial writing for AI authorship. (Pangram's methods are insensitive to dates, so I did not worry that the opinions' timing influenced these results.)
The results for 2026 looked very different. More than 50 opinions showed signs of AI authorship. The exact percentages ranged from less than 1% to more than 50% AI-written, with most results clustering near the bottom of that range. Still, the contrast with January 2022 was hard to miss.
We know judges are using AI. Should they be using it?
4 comments:
Yes, of course. The same rules go for them as attorneys or anyone else in any profession -- use AI or whatever tools help with efficiency in your trade, just make sure not to sacrifice the accuracy and quality of the output. With AI in the legal profession, that means personally and independently verifying every source and statement of fact, law, and conclusion the AI spits out. Straightforward rule to live by, and in my opinion there should be harsher penalties for attorneys, judges, and fiduciaries who can't abide by that principle.
My gut says no but my brain says yes. If AI produces truthful results, what is the difference between that and an engineer using a calculator or computer to build a bridge? Or better yet, think about it as an advanced way to use spell check.
Not sure whether they "should" be is a worthwhile question. The horse has left the stable and is galloping already. And what could we actually do? Lobby Congress to reduce their salaries? Ha!
Pretty sure that towards the end of anyone here's lifetime, we're already going to see judges, courts, and juries being replaced 100% by an LLM. As in, the parties (or their lawyers, if they decide they even need one) will just feed all the "facts" into some government-approved AI system, and within minutes or even seconds, the machine will spit out a binding "judgment."
It will of course start off being implemented in low-stakes cases--traffic court, small claims, etc. But, given the incredible cost-savings to local and state governments that they'll soon recognize in not having to pay judges, clerks, support staff, or maintain courthouses, etc., it won't take much longer before higher-stakes disputes get the same treatment.
Post a Comment