I looked over this as someone still dealing with things a person I have never met posted in my name 15 years ago. This really is bad for any side. A further rant I have put out there, the most useful precedent to test would be a "multi John Doe" restraining order, requiring a social media platform to delete and block reposting of identical harassing posts by one or a few users with an army of "sock" accounts without making the victim responsible for proving who they are.
A lawyer in our firm follows this sort of shit pretty closely. He tells me that, at present, there are just over 300 examples of lawyers getting wang-slapped in U.S. courts for filing AI-generated motions, briefs, etc. that include citations to imaginary case law.
The fact that generative AI makes shit up and includes the made-up shit in legal documents comes as no surprise. It just happens way too often to qualify as surprising anymore.
What STILL surprises me is that so many lolyers are so fucking lazy and/or incompetent that they can't be assed to check the citations in their AI-generated briefs to see whether the cited cases are, you know, real. Or even assign the jerb to someone else!
It's not just legal. AI is equally bad at doling out technical advice in the IT world. Makes up processes and commands that don't exist, discusses tables and fields that are completely imaginary. I'd say it ("it" in this case being the AI engine that Google uses when responding to search results) is at least partially wrong about 50% of the time and completely wrong about 10% of the time.
“It’s hard to “match” something like that when you are confined to reality.“ This, coincidentally, is why Democrats have an uphill fight against lyin’ Republicans.
So vaguely related, I was using AI to do some digging on some alt-country performers, writers, and lyrics. For personal use only, out of curiosity.
My AI tool first gave me erroneous information about a performer that I knew for certain had passed-- AI said he was still alive.
Second, it cited some data for me that confirmed one of my conjectures, so I asked for the link and followed it. I read every long page of where it sent me, and there was nothing there that corroborated my theory. AI was just telling me I was right and posting links to nothing. This shit is dangerous.
Same. When I still had a job, I tried using ChatGPT to speed up some software research, always prompting to cite sources. Half the time the links were hallucinated. And if you tell it that the link doesn't exist, the reply is, 'you're absolutely right.' Great use of water and power there.
My wife works in a library. The librarians are trying to figure out how to use AI, and whether or not it threatens their future employment.
One day over lunch, she did a quick search for all the candidates running for Governor here in Minnesota. Google's AI list included state Rep. Melissa Hortman as a probable contender.
Which she probably would have been, had she not been assassinated last June.
I looked over this as someone still dealing with things a person I have never met posted in my name 15 years ago. This really is bad for any side. A further rant I have put out there, the most useful precedent to test would be a "multi John Doe" restraining order, requiring a social media platform to delete and block reposting of identical harassing posts by one or a few users with an army of "sock" accounts without making the victim responsible for proving who they are.
Wait for the lawsuit Robyn.
Marctrent.ai single handedly proves that using AI is guaranteed to get your case thrown out of court. Here endeth the lesson.
Truly good news in an otherwise abysmal news season.
C'mon, lawyer guys, don't use AI -- just look everything up on Wikipedia and use that. No need to check those citations either.
AI ain't right, aight?
"It’s like having a crystal ball for your case." With the same amount of utility!
a broken crystal ball.
A lawyer in our firm follows this sort of shit pretty closely. He tells me that, at present, there are just over 300 examples of lawyers getting wang-slapped in U.S. courts for filing AI-generated motions, briefs, etc. that include citations to imaginary case law.
The fact that generative AI makes shit up and includes the made-up shit in legal documents comes as no surprise. It just happens way too often to qualify as surprising anymore.
What STILL surprises me is that so many lolyers are so fucking lazy and/or incompetent that they can't be assed to check the citations in their AI-generated briefs to see whether the cited cases are, you know, real. Or even assign the jerb to someone else!
It's not just legal. AI is equally bad at doling out technical advice in the IT world. Makes up processes and commands that don't exist, discusses tables and fields that are completely imaginary. I'd say it ("it" in this case being the AI engine that Google uses when responding to search results) is at least partially wrong about 50% of the time and completely wrong about 10% of the time.
How does someone end up being such a unmitigated douche?
"Efficiency on Steroids:" And we all know what steroids do to a guy.
Marc Trent is making Lionel Hutz look good.
Better not call Marc.
Fabulous. Also, shout-out to 404 Media, throw them bux if you can. Great tech-focused independent media source.
Second. The podcast is also so good!
“It’s hard to “match” something like that when you are confined to reality.“ This, coincidentally, is why Democrats have an uphill fight against lyin’ Republicans.
So vaguely related, I was using AI to do some digging on some alt-country performers, writers, and lyrics. For personal use only, out of curiosity.
My AI tool first gave me erroneous information about a performer that I knew for certain had passed-- AI said he was still alive.
Second, it cited some data for me that confirmed one of my conjectures, so I asked for the link and followed it. I read every long page of where it sent me, and there was nothing there that corroborated my theory. AI was just telling me I was right and posting links to nothing. This shit is dangerous.
Same. When I still had a job, I tried using ChatGPT to speed up some software research, always prompting to cite sources. Half the time the links were hallucinated. And if you tell it that the link doesn't exist, the reply is, 'you're absolutely right.' Great use of water and power there.
My wife works in a library. The librarians are trying to figure out how to use AI, and whether or not it threatens their future employment.
One day over lunch, she did a quick search for all the candidates running for Governor here in Minnesota. Google's AI list included state Rep. Melissa Hortman as a probable contender.
Which she probably would have been, had she not been assassinated last June.
My wife isn't worried about AI taking her job.
There was a movie about this with Spencer Tracy and Katherine Hepburn, called “Desk Set.”
And people wonder why I just can't be fucked to even attempt dating...
"his lawyer passionately argued that he was just literally too stupid to do his own taxes."
His lawyer got one thing right.
But not in a disparaging way, of course.
This creep and his circumstances remind me of the dating scene with background check in Amazon Women on the Moon: https://vimeo.com/437434336