Keith F. Lynch <
kfl@KeithLynch.net> wrote:
Getting back to AI, I wouldn't dignify the current chatbots with that
term. They're just slightly refurbished versions of the classic
"dissociated press" program of a half century ago. I laughed out
loud when I heard about the lawyer who trusted ChatGPT to do his job.
That "AI" didn't wipe out all humans, but it did wipe out the law
career of that one human.
More likely two humans. Their law firm may well also get dinged (over "supervision") but probably not enough to sink it.
IE, the original (State certified) lawyer was the one that used
ChatGPT to generate the original gibberish to the Federal court which
resulted in Avianca firing back "we can't find 5 of the cases and the
remaining two doesn't say what they claim they do".
The Judge clearly smelled something bad and ordered the other side to
"provide certified copies of all cases you are referencing" - a very
uncommon request which should be considered a major hint but wasn't.
At least one of the invalid references were to the Federal Reporter
which publishes all federal cases, it's just not possible to NOT find
a valid reference for that. Another reference mentions page 100,000+
something, clearly not physically possible, others were basically
gibberish.
Anyone SANE would instead have gone WTF have I done and immediately
fold but no, he asked ChatGPT if it was SURE... and then to provide
him with the cases, which it manufactured for him (it rarely say no),
and then he sent that in despite realizing they clearly wasn't the
full actual cases the Judge had requested because they were WAY too
short, the excuse was that he "thought" they were excerpts but that
was NOT what the Judge had asked for.
We know this because he submitted screenshots, which including the
text around the chat window clearly stating nothing it said could be
relied upon... Ouch.
Which leads us to attorney number two, the co-worker that took the
first ones work and sent them into the Federal court, due to the first
not being admitted to the Federal court and thus couldn't legally do
the work once Avianca (trivially) moved it from State to Federal court
system where it always was going to end up under the Montreal
Convention.
He didn't use ChatGPT *but* as the person sending them he's
RESPONSIBLE for anything he send in and he clearly didn't even read it
(there were red flags everywhere even on a cursory read-through!). But
he's also on hook for fradulent use of certification/attestion! and
outrightly lying to the Judge (when requesting an extension earlier he
claimed it was for him being on vacation, but it was really the other
lawyer that was on vacation).
I expect they're BOTH toast, albeit for different reasons. And really
not even by ChatGPT, I consider that more as an amplifier of stupidity
here.
For a non-expert it's really hard to know who gets it worse, both did
pretty much everything in their power to dig themself a deep hole.
And I wonder if there will be criminal reprecussions, it was a civil
case but we're basically talking "court fraud" (both) and "lying to a
judge in writing" (only the second).
I suspect there will be sanctions and referral to their Bar
Association(s) (that can disbarr them) but I'm not going to be
surprised if the Judge also referred to "investigation". And when a
Federal Judge do that it do get investigated and often lead to
criminal prosecution. Some of the things looks like they could lead to
jail time.
Honestly, it's one of those "you couldn't put this in a book because
no one would believe it".
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)