Page 21 - ACTL Journal_Sum24

we know what the rules are. And no matter what technology you use in order to submit a filing to the court, in order to make an argument to the jury, you know you have to comply with the rules.
So I do think we’re going to see the orders go away as people get smarter about it. I don’t think we’re going to actually see a proliferation. And if judges don’t understand generative AI themselves, then we’re going to get orders that prohibit the use of AI. “How do I comply? If I use a Word processor with spellcheck, is that running afoul of the judge’s order?”
Wolfsohn noted the Rule that obligates lawyers to have accu- rate, error-free briefs, whether AI generated or human gener- ated. Judge Thumma agreed that lawyers’ personal responsi- bility is no different whether they use AI or rely on any other traditional source for their work product.
Back for Pro’s last words – the opening statement it generated for the Menninger case. It wove emotions and storytelling into the theme it had crafted and created a narrative:
We are here today because of a tragedy that struck the Menninger family; a tragedy that was both preventable and forewarned. On a day like any other, Laura Men- ninger, a beloved member of our community and a devoted wife and mother, set out on her bicycle. She trusted that the roads in her community were safe and that the vehicles sharing those roads, including the au- tonomous vehicles touted for their safety, would protect her. That trust was misplaced.
Not bad.
Wolfe is a fan of AI. “Just dive in and start experimenting with this, keeping in mind the limitations for ethical reasons and what the judges are now telling us you can and can’t do. But go out there and experiment; just go wild. Give it a try because as you start playing with this and doing these prompts, this whole area of prompt engineering, I think that’s
the challenge we all face, that again, whatever we put into the system to cue it to do certain things, it’s going to give us that.”
Judge Thumma added that judges need to be educated about generative AI at a more profound level than “on the filing front.” Telling the difference between reality and deep fakes is very difficult, because AI is so good. His advice was to engage expert testimony to figure out what is and what isn’t reliable, and for counsel to know what foundation they need to show when offering evidence, such as photos. These are the upcom- ing challenges for AI at trial.
She added a comment on the issue of deep fakes and what is called the “liar’s dividend,” when somebody says, “How do you know it’s not a deep fake?” when it’s virtually unques- tionable that it’s real. How do you know it’s real? Be prepared for it and raise it with the court early. Don’t wait until the eve of trial. Bring it up really early so that there’s ample time to get discovery if you need it and for the court to give it a thoughtful response.
Fairless cautioned lawyers about compounding their AI mis- takes by lying to cover up their errors and attempting to hide the mistakes from the court. Don’t double down by being dishonest about it, trying to cover it up.
Fairless recommended reading Florida Ethic’s Opinion 24- 1, which addresses the many different ethical rules involved. Roslyn Levine wrapped up the Program. “If we asked AI what it thought about our Program, I think it might say, ‘You just touched the surface of what I can do. You need to have an- other session sometime soon.’ And that’s probably one of the
few truthful things it would say.”
Roslyn Levine, K.C. Toronto, ON
SUMMER 2024 JOURNAL 20
Professor Grossman distilled her essential advice to three words – trust but verify.