Judges Are Fed up With Lawyers Using AI That Hallucinate Court Cases
-
Yeah, I know how LLMs work, but still, if the definition of lying is giving some false absurd information knowing it is absurd you can definitely instruct an LLM to “lie”.
-
Cut the guy some slack. Instead of trying to put him in jail, bring AI front and center and try to use it in a methodical way...where does it help? How can this failure be prevented?
It can be prevented by people paid 400-1000 per hour spending time either writing own paperwork or paying others to actually write it.
-
I've had this lengthy discussion before. Some people define a lie as an untrue statement, while others additionally require intent to deceive.
The latter is the actual definition. Some people not knowing what words mean isnt an argument
-
Me: I want you to lie to me about something.
ChatGPT: Alright—did you know that Amazon originally started as a submarine sandwich delivery service before pivoting to books? Jeff Bezos realized that selling hoagies online wasn’t scalable, so he switched to literature instead.
Still not a lie still text that is statistically likely to fellow prior text produced by a model with no thought process that knows nothing
-
It can and will lie. It has admitted to doing so after I probed it long enough about the things it was telling me.
You can't ask it about itself because it has no internal model of self and is just basing any answer on data in its training set
-
Yeah, I know how LLMs work, but still, if the definition of lying is giving some false absurd information knowing it is absurd you can definitely instruct an LLM to “lie”.
A crucial part of your statement is that it knows that it's untrue, which it is incapable of. I would agree with you if it were actually capable of understanding.
-
All you do is a quick search on the case to see if it's real or not.
They bill enough each hour to get some interns to do this all day.
All you do is a quick search on the case to see if it’s real or not.
You could easily. We have resources such as LexusNexus or Westlaw which your firm should be paying for. Even searching on Google Scholar should be enough to verify. Stay away from Casetext though, it's new and mostly AI. LN and WL also have AI integration but it's not forced, you're still capable of doing your own research.
I've been telling people this for a while, but everyone needs to treat AI like how we used to treat the wiki. It's a good secondary source that can be used to find other more reliable sources, but it should never be used as your single standalone source.
I'm not going to sugarcoat it, AI is being forced everywhere you look and it is getting a bit difficult to get away from it, but it hasn't taken over everything to the point where there is no longer any personal responsibility. People need to have some common sense and double check everything as they've been taught to do even before AI.
-
I don't know if I would call it lying per-se, but yes I have seen instances of AI's being told not to use a specific tool and them using them anyways, Neuro-sama comes to mind. I think in those cases it is mostly the front end agreeing not to lie (as that is what it determines the operator would want to hear) but having no means to actually control the other functions going on.
Neurosama is a fun example but we dont really know the sauce vedal coocked up.
When i say proven i mean 32 page research paper specifically looking into it.
https://arxiv.org/abs/2407.12831
They found that even a model trained specifically on honesty will lie if it has an incentive.
-
Cut the guy some slack. Instead of trying to put him in jail, bring AI front and center and try to use it in a methodical way...where does it help? How can this failure be prevented?
LLMs are incapable of helping.
If he cannot find time to construct his own legal briefs, maybe he should use part of his money to hire an AGI (otherwise known as a human) to help him. -
But the explanation and Ramirez’s promise to educate himself on the use of AI wasn’t enough, and the judge chided him for not doing his research before filing. “It is abundantly clear that Mr. Ramirez did not make the requisite reasonable inquiry into the law. Had he expended even minimal effort to do so, he would have discovered that the AI-generated cases do not exist. That the AI-generated excerpts appeared valid to Mr. Ramirez does not relieve him of his duty to conduct a reasonable inquiry,” Judge Dinsmore continued, before recommending that Ramirez be sanctioned for $15,000.
Falling victim to this a year or more after the first guy made headlines for the same is just stupidity.
I’m all for lawyers using AI, but that’s because I’m also all for them getting punished for every single incorrect thing they bring forward if they do not verify.
-
I've had this lengthy discussion before. Some people define a lie as an untrue statement, while others additionally require intent to deceive.
You can specifically tell an ai to lie and deceive though, and it will…
-
Still not a lie still text that is statistically likely to fellow prior text produced by a model with no thought process that knows nothing
Lie falsehood, untrue statement, while intent is important in a human not so much in a computer which, if we are saying can not lie also can not tell the truth
-
No probably about it, it definitely can't lie. Lying requires knowledge and intent, and GPTs are just text generators that have neither.
So it can not tell the truth either
-
So it can not tell the truth either
not really no.
They are statistical that use heuristics to output what is most likely to follow the input you give itThey are in essence mimicking their training data
-
not really no.
They are statistical that use heuristics to output what is most likely to follow the input you give itThey are in essence mimicking their training data
So I think this whole thing about whether it can lie or not is just semantics then no?
-
So I think this whole thing about whether it can lie or not is just semantics then no?
everything is semantics.
Lying is telling a falsehood intentionally
LLM's clearly lack the prerequisite intentionality
-
everything is semantics.
Lying is telling a falsehood intentionally
LLM's clearly lack the prerequisite intentionality
They can’t have intent, no?
-
Me, too. But it also means when some people say "that's a lie" they're not accusing you of anything, just remarking you're wrong. And that can lead to misunderstandings.
Yep. Those people are obviously "liars," since they are using an uncommon colloquial definition.
-
Haven't people already been disbarred over this? Turning in unvetted AI slop should get you fired from any job.
I heard turning in AI Slop worked out pretty well for Arcane Season 2 writers.
-
you sound like those republicans that mocked global warming when it snowed in Texas.
sure, won't take your job today. in a decade? probably.
Going off the math and charts that OpenAI and DeepMind both published before the AI boom which correctly guessed performance to cost ratios: we've reached the peak of current models. AI is bust, mate. In particular, Deepmind concluded with infinite resources the models in use would never reach accurate human language capabilities.
You can say stuff like "they'll just make new models, then!" but it doesn't really work like that, the current models aren't even new in the slightest it's just the first time we've gotten people together to feed them power and data like logs into a woodchipper.
