I don't think you know what Plagiarism is. AI does not plagerize. It reasearches and uses that research, just like every historian that has ever wrote a book. It does not give direct quotes without citation
Ok, who gets standing: the company paying the electric bill or the machines? Maybe something really interesting is about to happen like... companies are no longer able to build a portfolio of intellectual property because ai gets to the patent office first. There goes stock options: full corporate meltdown, because whats the diff between one legal fiction and the other?
Before AI, I checked the citations in an opponent's brief, and although the cases existed, I found they did not support the position my opponent had claimed that they did. Lesson - always check citations 😂
Before generative pretrained transformers you mean? (We've used "AI" to describe just about every simulation of intelligence. Consider video games from the 1990s.)
Common for lawyers to use cases that don't actually support a proposition. Depends on how it's used. But when I clerked for a federal judge, I often found citations to cases (always read the cases cited) were not just nonsupportive but when read fairly actually supported the opponents side. I loved citing those cases in drafting the judge's opinion.
@@YadraVoat reminds me of what a manager of mine told me about his thesis. he invented a ukrainian person, who just so happened to have the same name, only backwards, and quoted his papers, which conveniently were only released in ukrainian. back in the 1970ties, nobody had a chance to check that stuff. oh, and the citations always supported his own findings.
I can't imagine this "expert" ever having credibility in front of any court. If he argued a parking ticket, I'd use this to establish he has a pattern of dishonesty.
He was tapped as an expert specifically in how the existence and use of AI harms trust. Task failed successfully! Unfortunately, ticking off the judge overrides how well he accidentally proved his point. All because he got lazy and had it do citations, the same thing that has previously bitten lawyers trying to shortcut their work. And now the case is practically guaranteed to succeed, which means campaigning candidates in MN will be able to freely use AI-generated deepfakes of their opponent.
Ai song generation hallucinations can get creepy. Its always funny when the Ai starts to hallucinate halfway through a song or after the song is over... but sometimes it just gets soooo dark and creepy with the voices... lol
Computer programs, including AI, don't have the ability to randomize anything. Calling them hallucinations is in itself, a lie by whoever coined the term.
Using tech, doesn't make you smart. Its like thinking going to the drive thru window, makes you smarter than a hunter/gatherer. To the contrary, the modern high tech world makes you stupid !
The Professor tried to prove AI was dangerous by having ChatGPT write the brief... But all it generated were recipes for making radio transmitters out of coconuts and turning Ginger's evening gowns into solar panels!
I can't tell if this is a very clever statement or not. I'm going to assume it's clever. There is no I, but an l can look like I just as LLMs look like intelligences.
He should be sanctioned he committed perjury by filing the document without first reviewing it to confirm the accuracy of its contents. I don't feel sorry for attorney's that ignore process, because process matters.
@@briggsc4 I highly doubt the commenter is saying the attorney doesn't deserve the sanctions or feels sorry for them. Rule 11 sanctions are just such A serious and potentially career ending matter that any attorney hearing it has A visceral reaction. The context of your comment makes it seem like you are implying that the commenter disagrees with what you are saying. If that isn't the case my bad, I misunderstood what you meant to say.
I am in law school atm, in part thanks to a video by Lehto inspiring me, and we are currently learning about AI in class, and specifically, about rule 11. yeah this isnt good for them lmao
I think of it like The Terminator, except there's no human resistance in sight. Humans are all too eager to let the machines take over, starting with their jobs.
It should be on the AG of Minnesota to bring a criminal case on his own expert just to make the point, but the AG will probably have to appoint an outside counsel because it's his own witness.
My insurance company kept sending drones to snoop on my backyard, so I decided to use AI to help me file a complaint, but Honey intercepted my message and instead changed it to a purchase order for a used car from Ali Express. AE replied by mailing me a picture of a rental car from Hertz. 😒
Does that chatAi app have the ability to create something resembling music? Michael Smith is a man based in North Carolina who has been charged with wire fraud and money laundering for creating millions of AI generated songs, uploading them to streaming platforms like Spotify and then using bots to "listen" to those songs in order to get royalty payments totaling over 10 million dollars. Does anyone know if Steve has covered this story? If not, who else would like to see him do a video about it?
Because it isn't worth the time and money to do it. It wasn't material to the case in fact or purpose. It was only the citations that got screwed up according to the professor.
Because there was a quick response, a show of remorse, and a lack of motive beyond being lazy. I commented on another channel that made a video about the same case that the court was (necessarily) lenient here and l this should definitely be something that sparks an investigation into his academic works (a thing Stanford is getting used to doing at this point), but an actual perjury charge would be a waste of taxpayer resources.
In the academic evolution of responsibility we have gone from the dog ate my homework to the dog drafted my homework, you can't hold responsible or blame me for submitting it without verifying it. The dog is man's best friend after all.
He's a hero honestly. Sounds like the court finally understands the perils of AI and are taking action. The courts are inept if they are just figuring this all out now, the professor knew that, and made a huge sacrifice for all of us. Proffessor 1, court 0. He lost his career, but he was probably about to retire anyways.
I LOVE using AI when engaged in arguments, but despite these arguments having no significant consequences attached to them, I STILL doublecheck all the information the AI finds for me, because I don't like looking like an idiot. If I was having to use that information in court, I would be fact checking the hell out of it!
I would think so, wouldn't this classify as basically knowingly and intentionally using false information that could sway a case into a specific direction?
Probably depends when it's discovered. The attorneys are responsible to make note of anything that is amiss enough for an appeal. If they could have known during the trial, (for instance if they could have but didn't look at the citations) but fail to object, they might not be able to bring it up later.
I would hope a judge is not just blatantly believing everything a lawyer says without verifying the cited case law. But this is America... land of the lazy.
A heavy factor would be, how influential this expert's testimony is believed to have been. Complicating factor: Courts are completely blind to jury deliberations.
Wikipedia has been around for quite a while, and I don't trust it for legal matters. So in what universe is an attorney trusting AI at this early stage already? 😢
There's a lawyer in England, his TH-cam channel is BlackBeltBarrister, who has set up an A.I. to answer people's questions on things like traffic tickets, to find out if they have a proven legal case (case law) to argue or should just pay. I guess that if you have a V.P.N. to appear from England, you can try it out with scenarios.
Had to jump on my computer to comment on this one. This was the funniest video I've seen in a long time. No apologies for length. Always appreciate your entertainment and humor. Thank you for keeping it real.
I got an extreme story Im being sued out of my home for not paying property taxes on a fake address on a made up account .. I cant get anyone to care enough to even write the story
Local news station that has one of those “BRING IT TO THE PUBLIC” sections would not only help you, but bring forward anyone else they have targeted. Now, the big thing is WHO did you tick off? That can have an effect on an “Abuse of Power” charge on not only that person, but anyone that HELPED them!
@@RatKindler John McCarthy? He was a computer scientist who had major computer science contributions including programming language design. Its possible to have multiple expertise but there isn't any doubt that he had a strong mathematics and computer science background. While the professor in question here does not.
@@username7763 I'm referring to Geoffrey Hinton. The field of AI is not strictly computer science and neural networks though. It also includes how it models biology as well as how it's used and its impact on individual behaviour and society. An "expert" in AI doesn't mean their knowledge is only in computer science - just one aspect of the field.
I have a friend that is a lawyer and he loves this kind of thing too. The biggest Perry Mason moment I have seen in recent times was in the Baldwin trial. Examination of the lead prosecutor was the most entertaining thing ever. Her credibility as a lawyer was destroyed. Why she ever got on the stand in the first place is beyond me. She knew at that point her goose was cooked she might as well have just left with her dignity somewhat intact.
He did the research and wrote the arguments himself. He used AI assistance to rephrase his sentences to improve the quality of the writing. Unfortunately it also replaced his "insert citation here" markers with, umm, what it said he wanted there. He didn't notice because they looked the same as the legitimate citations he had placed elsewhere in the document. Leonard French made a video a couple of weeks ago where he read out the actual court filing. If I recall, one of the incorrect citations was for a document he himself published on the subject (being someone who works in the field), which got replaced with someone else's name. All of the things he referenced were real documents, the citations were just wrong. Quite frankly as a regular viewer I am disappointed at Steve's editorialising on selected snippets of a case and misrepresenting what happened, when the detailed explanation of what happened is available to read for yourself.
Mr. Lehto’s videos are usually quite easy to follow. This one, at 3:20 in, ranks up there with the poison scene with Danny Kay from “The Court Jester”. Don’t attempt it after a drink or two. I’m going to try again later, when completely sober. I think it may become my favorite of all time.
If he ever gets that invite, he probably shouldn't accept it. It might be hard to maintain a court room appropriate demeanor if he did accept it. Steve would be having so much fun cross examining that so-called expert that the smile on his face would probably get him in trouble with the court.
Since ellison is partly responsible, since he presented the affidavit to the court as evidence, he should probably be disbarred and removed from office.
And you can be totally sure that's not just partisan bias. It's not like the comment section indicates a fair percentage of the opposition party wrote a comment saying he's unethical and corrupt - without a single one of them naming an example. The politicians are a symptom, the sickness is people who call them guilty or innocent based solely on the letter after their name.
If you "make something up to prove a point," it means you are creating a fictional scenario or example, often exaggerated, to illustrate your argument and convince someone of your perspective, even if the specific details aren't entirely true; essentially, using a hypothetical situation to strengthen your point. Key points about "making something up to prove a point": • Not meant to be literal: The fabricated story or example is not intended to be taken as factual, but rather as a tool to explain a concept or idea more clearly. • Can be risky: While useful in certain situations, using made-up stories can sometimes be seen as manipulative or dishonest if not used carefully. • Important to clarify: If you are using a hypothetical example, it's crucial to explicitly state that it's fictional and not meant to represent a real situation. Example: • "Imagine a world where everyone only wore one color of clothing - this would clearly demonstrate how diversity can add richness to society." Generative AI is experimental.
If the purpose of the expert witness was about the risks of misinformation by AI, wouldn't the AG actually want to keep that affidavit, but turn it into evidence of AI misinformation? It just seems to me this is a case of the worst possible action can be used for the best possible outcome.
@@Bobs-Wrigles5555 This was one of his sneakier placements. I am good at checking inside the handles...and on top of the microphone! Missing that one really hurt.
Excluding punctuation, I proofread my own comments before posting. It's not 100% foolproof but it keeps me from posting something I would probably regret later.
It’s also worth pointing out that he never had any authority to claim to be an expert in the first place. None of his qualifications relate to AI or AI safety. He is an internet archivist. It is outrageous that he was allowed into the court at all.
"Deep Fakes?" Like trying to convince a court of law that you've written something you haven't, and submitting an affidavit, including citations that don't exist?
I've wondered the same thing. Is it being fed (programmed) with bad information to begin with...or is it taking a life of its own and crating alternative facts with said information. Next thing you know it will say you can divide by zero!
It didn't. The AI is familiar with how citations should exist in a certain type of document, and when it's given a prompt to recarate a document of that type, it's going to essentially copy that format with what it creates. Problem is an AI doesn't know what's true or what's not, it's just putting characters onto the page to make The document it generates look like it should. It doesn't know it's something that doesn't exist, because it didn't look up the citation. It just scraped the internet for information related to the prompt, and put text on a page in the right format.
Our current ai is not programmed. It's just a fancy autocomplete. It has no concept of facts, knowledge or anything. All it does is predict the new characters
AI is a misnomer it's machine learning. It reads a ton and paraphrases it back to you. So if it read false info it leaks into it's results. A lot of these models were just given access to opinion websites like reddit, so they have junk in junk out. That's why good says it's good to eat several small rocks a day, because it was a reddit comment, a joke. But the "AI" has no actual learning elements it just generates text. Real AI has to go through hundreds of thousands of iterations with punishment/reward parameters. Can't do that for these kinds of chatbots because you need a human to fact check every iterations. It's fancy auto complete and that is all it will ever be. It's all a ponzi scheme to milk money out of investors and the stock market.
I'm Pro Se, I use gpt to write drafts, but I check all citations including the actual verbage in the case. Gpt still makes up case law all the time....
Whether 'thrown out and resubmission not allowed' or 'career ruined'... Expert witnesses and attorneys (future judges) should consistenly suffer the consequences prescribed by law... else the importance of justice and of following the law appears as arbitrary (often as weighted in favor of the professional/privileged class and/or those in positions of wealth or power). This seems to be one of the things that has allowed (or is simply another example of) corruption to permeate every level of our society.
As a Minnesotan, the sad part is that there is nothing about this that surprises me. It's pretty much a case of " Stupidity As Normal." But, in Ellison's defence, he has to submit breifs with Artificial Intelligence in them. It's my understanding that there needs to be some sort of intelligence in court proceedings, and he doesn't have any Real Intelligence of his own.
FYI, if using a generative LLM (such as chatGPT) preface your question with "Do not make up responses." and the LLM will produce fewer fictious answers.
You may get better results if you include the citation documents as attachments (inline or RAG) to the prompt. But you still need to double check the result.
@@DVSProductions Try it both ways. By explicitly adding a prohibition against making up responses you are creating a "punishment" condition that doesn't exist by default.
*I think Perjury is a Crime! What is the penalty for Perjury? I will bet the "Expert" will not suffer the penalty for Perjury!* *He should now be a Professor in the State Prison! 2024 Minnesota Statutes - 609.48 PERJURY. Just-Us!*
@@monkemode8128 Didn't he intentionally do that though? If he would have had some other person write that for him and then passed it on as his own without checking it, would that be any different? "No I did not kill that man, I only told paid that other man to do it."
@ From my foreign point of view, there's insane amount of bizzare scapegoating happening in USA like it appears to be in this case too. People who are obviously responsible being let go because someone lower down in the pecking order is put out as a scape goat to take the blame or something non human is to blame. I do not have the words to express myself how insane this appears to me.
The expert's biggest mistake was failing to verify the citations. But, once they were found to be false, that called into question the supposed facts stated in the declaration. 😮
You said what will be the downfall of the modern civilization. Critical thinking skills. It is not taught anymore. Children are taught what to think, not how to think. Now couple this whith a short attention span and the desire for immediate gratification.
The primary cause of being taught "what to think, not how to think" was the automation of education via machine-graded multiple choice tests, bring things back around to lazy, blind misuse of computers.
Truly, this one sounds like it could have been written by The Onion. Though, this (admittedly wonderfully ironic) example of hazards of using AI in a court case about the hazards of AI pales before The Onion’s use of parody in that legendary amicus brief. (:
Minor point: the software that the professor uses for academic papers is probably a citation manager which won’t have a baked in function to double check chatGPT’s work. They let you import full citations from a known good database (such as pubmed) and format them to the style desired by the journal you submit to. They can’t check an arbitrary exported list of citations. However, it could be possible to have ChatGPT generate a list of unique identifiers for its citations (ie ISBN, PMID) and then manually check them either in the citation manager or something like Google scholar. That is more work than “trust me bro” obviously.
That's another AI snafu that Leonard French made a video about, a couple of days ago. Since Co-Pilot is part of 365 (what they renamed Office), and lawyers and courts use Word documents, which is part of 365/Office. Co-Pilot risks the confidentiality of all those documents !
@@michaelmoorrees3585 I was using the one built into bing thing.... I asked it who was using copilot and actually liked it. The AI didn't really answer my question but it asked me if I plan on using copilot and I answered "No, I abore the idiea of copilot and it creeps me out". The AI said it understood that copilot is obtrusive or something to that nature... and many people are not in favor of it.
I work for a religious organization this reminds me of an Ethical AI Use class. "AI can be a powerful tool but it cannot and should not be what is delivered by messaging. Like any task you are the steward of the work and are responsible for everything written "
My dad worked as a technician in an oil refinery when I was growing up. One part of his job was sighing off on equipment that needed work. Usually such work was performed by outside contractors who may or (more usually) may not be failure with the operation of such equipment. There are a great many things you can do in an oil refinery that can make you and others unalive very quickly. For this reason, when such work is needed a technician needs to make sure the equipment is safe to work on. He never delegated this work to others. If he was signing off on it, he wanted to make sure it was, in fact, safe to work on. The consequences, go well beyond being responsible for killing people, if he signed off that something was safe when it wasn't could mean prison. An accident took place in a separate refinery to a pipe that carried molten sulfur was empty, not under pressure and safe. It was not. The contractors opened the line and several tons of molten sulfur sprayed out and killed 15 people. Almost the entire staff of a particularly good contractor. I believe the technician who signed off the equipment was safe spent the rest of his life in prison.
Little pro tip for anyone using AI to write anything. Do NOT copy-paste the output. Re-type the whole thing trying to rephrase things in the way that you normally would where it makes sense to do so. Some times some ideas are so simple that there is only one way to express it without overcomplicating it for the sake of saying it differently. That is outside of the obvious things like following every single thing that it links to as a source and making triple sure that the reference exists, is correct, is reliable, and that the reference is being done correctly. I should say that I thought those were so obvious that they were not even worth mentioning, but clearly not....
Speaking of billable hours, does anyone remember the WPPSS debacle in the PNW? And how the lawyers were flying across country to bill more hours each day they worked on the lawsuits?
I'm an attorney in NY, and about once a year I catch another attorney using a block quote that omits words like "not" to alter the court's determination in their cite.
I heard about this case. 🤣🤣🤣 The AG deserves this. "Are you saying we need to ask our experts if they use AI and how?" YES. That is exactly what we are saying.
Easy enough, the court should have it's clerk run all filings through AI to verify citations. I bet Chat GPT is pretty accurate at verifying citations as opposed to finding an important relevant citation in context of the legal theory you are pursuing.
@TheRealScooterGuy traffic lights 🚦 I'm tired of waiting for non-existent cross traffic. Cameras can look for cars and trigger the sequence if none are coming. Could also trigger all red when an emergency vehicle is coming or clear the intersection if the intersection is clogged.
@@tony_25or6to4 -- Agreed. I've said much the same elsewhere. But don't limit it to standalone management of one intersection. Let one intersection tell the next what is coming, and when. Let the system notice that construction has affected the traffic, and adjust accordingly. Teach it to recognize what happens with traffic when the weather changes. Also, give it multiple cameras and sensors watching overlapping sections of roadways, and recognize when one sensor or camera is sending false data, so that is can be scheduled for maintenance. But whatever we do, don't let it write legal briefs until it can tell the difference between truth and falsehood in its own writing.
The more articles, legal cases, and lectures I see by college professors and so-called learned experts, I mourn for the educational system. With a few exceptions, most of them seem to be full of it.
The funny thing about all this is the AIs will straight up tell you that they hallucinate... I asked CoPilot about hallucinations and how to correct them. It said that it does not know how to correct itself, so usually it just fills in the blanks to make the item look more complete, and relies on people to review what it writes for accuracy. When I objected that that was dishonest and unethical, it said that it did not understand ethics.
You know something that scares me is I've been hearing recently on the news how they're trying to put AI into hospitals and healthcare and diagnosing stuff. It scares me because of all these mistakes that AI makes
Waiting for an AI suing another AI for plagiarizing their own plagiarism
That's probably coming up soon! Open AI has started saying DeepSeek used their algorithm~.
That's going to be a great day. Not by judicial standards, necessarily, but the world will be laughing.
I don't think you know what Plagiarism is. AI does not plagerize. It reasearches and uses that research, just like every historian that has ever wrote a book. It does not give direct quotes without citation
@@Justise lol yes it does
Ok, who gets standing: the company paying the electric bill or the machines? Maybe something really interesting is about to happen like... companies are no longer able to build a portfolio of intellectual property because ai gets to the patent office first. There goes stock options: full corporate meltdown, because whats the diff between one legal fiction and the other?
Before AI, I checked the citations in an opponent's brief, and although the cases existed, I found they did not support the position my opponent had claimed that they did. Lesson - always check citations 😂
Before generative pretrained transformers you mean? (We've used "AI" to describe just about every simulation of intelligence. Consider video games from the 1990s.)
Good point about how humans can cite real cases just as uselessly as GPTs can usefully cite phantasmal cases, though! 🤣
Please, hammer, don't hurt them 😂
Common for lawyers to use cases that don't actually support a proposition. Depends on how it's used. But when I clerked for a federal judge, I often found citations to cases (always read the cases cited) were not just nonsupportive but when read fairly actually supported the opponents side. I loved citing those cases in drafting the judge's opinion.
@@YadraVoat reminds me of what a manager of mine told me about his thesis. he invented a ukrainian person, who just so happened to have the same name, only backwards, and quoted his papers, which conveniently were only released in ukrainian. back in the 1970ties, nobody had a chance to check that stuff.
oh, and the citations always supported his own findings.
I can't imagine this "expert" ever having credibility in front of any court. If he argued a parking ticket, I'd use this to establish he has a pattern of dishonesty.
He was tapped as an expert specifically in how the existence and use of AI harms trust. Task failed successfully!
Unfortunately, ticking off the judge overrides how well he accidentally proved his point. All because he got lazy and had it do citations, the same thing that has previously bitten lawyers trying to shortcut their work. And now the case is practically guaranteed to succeed, which means campaigning candidates in MN will be able to freely use AI-generated deepfakes of their opponent.
I cannot express how I love the term 'hallucinations'.
It's a euphemism for "lies."
Under certain agreeable conditions hallucinations can be fun 🙂
Ai song generation hallucinations can get creepy. Its always funny when the Ai starts to hallucinate halfway through a song or after the song is over... but sometimes it just gets soooo dark and creepy with the voices... lol
Computer programs, including AI, don't have the ability to randomize anything. Calling them hallucinations is in itself, a lie by whoever coined the term.
@@johnbriggs3916 A lie requires intent. An AI has no intent. Hallucination is the technical term.
The smarter we become in technology, the stupider we become in common sense.
“Computers- Making our lives easier one more problem at a time” - Me.
Using tech, doesn't make you smart. Its like thinking going to the drive thru window, makes you smarter than a hunter/gatherer. To the contrary, the modern high tech world makes you stupid !
@@RajaniIsa
I call computers and microcontrollers "confusers" 😂
It's Cause and Effect.
Who do you think is programming that stuff ?
Thank you. 💯
The Professor tried to prove AI was dangerous by having ChatGPT write the brief...
But all it generated were recipes for making radio transmitters out of coconuts and turning Ginger's evening gowns into solar panels!
And STILL couldn't fix a boat .
I LOVE this comment!
Remember, the "i" in "LLM" stands for intelligence
Say, what? I can't find the I in LLM. Is this like the Fxxx in Falstaff? 🤔
Amusingly in most sans-serif typefaces, the capital-i is indistinguishable from lower-case L.
I can't tell if this is a very clever statement or not. I'm going to assume it's clever. There is no I, but an l can look like I just as LLMs look like intelligences.
😂
I prefer to call it artificial stupidity. In this case, combined with the natural stupidity of the State AG.
You mentioned "Rule 11," I said (out loud) "oh, sh!7, Rule Eleven!" and then you said "every attorney just got chills." 🤣
Sounds kind of like fight club rules, or for Dragon Ball Z Abridged fans, the rules of Popo’s training (and the PTSD they inspire).
He should be sanctioned he committed perjury by filing the document without first reviewing it to confirm the accuracy of its contents. I don't feel sorry for attorney's that ignore process, because process matters.
@@briggsc4 I highly doubt the commenter is saying the attorney doesn't deserve the sanctions or feels sorry for them. Rule 11 sanctions are just such A serious and potentially career ending matter that any attorney hearing it has A visceral reaction. The context of your comment makes it seem like you are implying that the commenter disagrees with what you are saying. If that isn't the case my bad, I misunderstood what you meant to say.
I am in law school atm, in part thanks to a video by Lehto inspiring me, and we are currently learning about AI in class, and specifically, about rule 11.
yeah this isnt good for them lmao
Is there any doubt that we're now in the age of idiocracy?
The age of Idiocracy propagated by a particular individual NOT in jail. SMH
I think of it like The Terminator, except there's no human resistance in sight. Humans are all too eager to let the machines take over, starting with their jobs.
I'm ready to vote for Terry Cruz
@@glarynth Joseph Jaquard and the Banū Mūsā approve of your message.
@@gregbillman42whom?
Thank you Steve, for saying "generative" instead of simply "AI" (many commenters are neglecting to include the operative word).
The only AI that I trust is that developed by Dr. Fastolfe, R. Daniel Olivaw. He is a main character in Isaac Asimov's novels, the R stands for robot.
It should be on the AG of Minnesota to bring a criminal case on his own expert just to make the point, but the AG will probably have to appoint an outside counsel because it's his own witness.
Minnesota won’t heal until the woke politicians are removed from office.
Ellison is not a person of integrity unfortunately. I have never voted for him.
When you decide to copy someone else's homework but their the dumbest kid in class.😅
I see what you did they’re … and I approve. 😅🤣
Was the use of the wrong ‘their’ intentional to make your point? It should have been they’re
“ their “ hahaha
I don’t se how that would “effect” the outcome, lol!
You is using their wrong.
My insurance company kept sending drones to snoop on my backyard, so I decided to use AI to help me file a complaint, but Honey intercepted my message and instead changed it to a purchase order for a used car from Ali Express. AE replied by mailing me a picture of a rental car from Hertz. 😒
Does ChatGPT have a Fifth Amendment right to remain silent?
possibly... the ability to exercise it? Not a chance lol
Does that chatAi app have the ability to create something resembling music? Michael Smith is a man based in North Carolina who has been charged with wire fraud and money laundering for creating millions of AI generated songs, uploading them to streaming platforms like Spotify and then using bots to "listen" to those songs in order to get royalty payments totaling over 10 million dollars. Does anyone know if Steve has covered this story? If not, who else would like to see him do a video about it?
Corporations Are People !
Yes. But It doesn't have the ability...😅
No, corporations have some rights yes, but not the 5th.
Why is an expert witness who commits perjury not up on charges for perjury ?
Because it isn't worth the time and money to do it. It wasn't material to the case in fact or purpose. It was only the citations that got screwed up according to the professor.
Because there was a quick response, a show of remorse, and a lack of motive beyond being lazy. I commented on another channel that made a video about the same case that the court was (necessarily) lenient here and l this should definitely be something that sparks an investigation into his academic works (a thing Stanford is getting used to doing at this point), but an actual perjury charge would be a waste of taxpayer resources.
The threats the courts levy are only viable against the peons.
In the academic evolution of responsibility we have gone from the dog ate my homework to the dog drafted my homework, you can't hold responsible or blame me for submitting it without verifying it. The dog is man's best friend after all.
🤣🤣🤣
But how do you explain to the dog that your homework ate the teacher?
He's a hero honestly. Sounds like the court finally understands the perils of AI and are taking action. The courts are inept if they are just figuring this all out now, the professor knew that, and made a huge sacrifice for all of us. Proffessor 1, court 0. He lost his career, but he was probably about to retire anyways.
I LOVE using AI when engaged in arguments, but despite these arguments having no significant consequences attached to them, I STILL doublecheck all the information the AI finds for me, because I don't like looking like an idiot. If I was having to use that information in court, I would be fact checking the hell out of it!
Retroactively discovering that made up citations were included in a completed case will make an interesting appeals process. Is it even appealable?
I would think so, wouldn't this classify as basically knowingly and intentionally using false information that could sway a case into a specific direction?
Probably depends when it's discovered. The attorneys are responsible to make note of anything that is amiss enough for an appeal. If they could have known during the trial, (for instance if they could have but didn't look at the citations) but fail to object, they might not be able to bring it up later.
@@brianorca Not a lawyer, but couldn't you argue ineffective assistance of council in that case? Seems like something a lawyer should catch.
I would hope a judge is not just blatantly believing everything a lawyer says without verifying the cited case law. But this is America... land of the lazy.
A heavy factor would be, how influential this expert's testimony is believed to have been. Complicating factor: Courts are completely blind to jury deliberations.
Wikipedia has been around for quite a while, and I don't trust it for legal matters. So in what universe is an attorney trusting AI at this early stage already? 😢
Wikipedia articles generally cite source data. You need to verify that source.
In the universe where a lot of people feel overworked and want the computer to create them something in 30 seconds, that would have taken them 6 hours
@@zacharyhenderson2902 But they still want to bill their client for the full 6 hours...
There's a lawyer in England, his TH-cam channel is BlackBeltBarrister, who has set up an A.I. to answer people's questions on things like traffic tickets, to find out if they have a proven legal case (case law) to argue or should just pay.
I guess that if you have a V.P.N. to appear from England, you can try it out with scenarios.
Steve, I appreciate your throwback to a slower, gentler time @ 21:48
"My video tape will run out." 🤣
Laziness will someday bring down our country .
Had to jump on my computer to comment on this one. This was the funniest video I've seen in a long time. No apologies for length. Always appreciate your entertainment and humor. Thank you for keeping it real.
The irony is the type of case this was. But in a sneaky, weird way, the affidavit may help prove something about the pitfalls of AI.
I got an extreme story Im being sued out of my home for not paying property taxes on a fake address on a made up account .. I cant get anyone to care enough to even write the story
Wow
Local news station that has one of those “BRING IT TO THE PUBLIC” sections would not only help you, but bring forward anyone else they have targeted.
Now, the big thing is WHO did you tick off?
That can have an effect on an “Abuse of Power” charge on not only that person, but anyone that HELPED them!
Dude, that sucks, I lament that you must deal with the ideocracy of others. I would however suggest that you contact the (Institute for justice)
I would think they’d foreclose on the fake property.
Learn how to weld. Buy a bulldozer. 😂
Steve,
I looked up the story; the Stanford professor isn’t an AI expert. He is a communication and social science professor.
…so the”expert” committed fraud…but being a social scientist, his reputation will not suffer at all.
He can still be an AI expert and be a social science professor. The "father" of AI who won the Nobel prize for it was a psychologist.
He is probably just some left wing buddy of Ellison who the ag thought would add credibility to the ag case
@@RatKindler John McCarthy? He was a computer scientist who had major computer science contributions including programming language design. Its possible to have multiple expertise but there isn't any doubt that he had a strong mathematics and computer science background. While the professor in question here does not.
@@username7763 I'm referring to Geoffrey Hinton. The field of AI is not strictly computer science and neural networks though. It also includes how it models biology as well as how it's used and its impact on individual behaviour and society. An "expert" in AI doesn't mean their knowledge is only in computer science - just one aspect of the field.
OMG this sounds like an article from The Onion. XD
It might eventually become an article in The Onion and then AI will cite it.
More likely the babylon bee
The controversy of AI in court highlights the devaluation of truth and blind acceptance of statements in public discourse.
I have a friend that is a lawyer and he loves this kind of thing too. The biggest Perry Mason moment I have seen in recent times was in the Baldwin trial. Examination of the lead prosecutor was the most entertaining thing ever. Her credibility as a lawyer was destroyed. Why she ever got on the stand in the first place is beyond me. She knew at that point her goose was cooked she might as well have just left with her dignity somewhat intact.
My guess is that there is also a fraud happening. Use AI but then charge the client for the extensive "research" that is done.
He did the research and wrote the arguments himself. He used AI assistance to rephrase his sentences to improve the quality of the writing. Unfortunately it also replaced his "insert citation here" markers with, umm, what it said he wanted there. He didn't notice because they looked the same as the legitimate citations he had placed elsewhere in the document.
Leonard French made a video a couple of weeks ago where he read out the actual court filing. If I recall, one of the incorrect citations was for a document he himself published on the subject (being someone who works in the field), which got replaced with someone else's name. All of the things he referenced were real documents, the citations were just wrong.
Quite frankly as a regular viewer I am disappointed at Steve's editorialising on selected snippets of a case and misrepresenting what happened, when the detailed explanation of what happened is available to read for yourself.
So in reality we should accept that AI is the best expert to testify on the dangers of AI? Seems legit.
Sure looks like he proved his point on the dangers. If a paid expert can’t do it, what about the general public.
Ben is hugging "Betsy's" hilt.
Where did that name come from??? Good afternoon Idris.
"Betsy", Was that in the set tour that I haven't got to watch yet?
Morning Idris
@@BenLeitch Good afternoon. I think it was a Steve original phrase.
@@Bobs-Wrigles5555 Good afternoon. I don't recall Betsy being mentioned in the last set tour but I know Steve has been using the name for a long time.
@@idristaylor5093 You have a better memory than me Sir, and Good Afternoon.
Good old Ellison is known to be a joke by the entire state of Minnesota.
100%, total dei hire!
Mr. Lehto’s videos are usually quite easy to follow. This one, at 3:20 in, ranks up there with the poison scene with Danny Kay from “The Court Jester”. Don’t attempt it after a drink or two. I’m going to try again later, when completely sober. I think it may become my favorite of all time.
Oh, the humanity of it all!!
If you ever receive that invite, please let us know. I'd love to come watch. 🤣🤣 🍿
If he ever gets that invite, he probably shouldn't accept it. It might be hard to maintain a court room appropriate demeanor if he did accept it. Steve would be having so much fun cross examining that so-called expert that the smile on his face would probably get him in trouble with the court.
Keith Ellison was always incompetent and a horrible human being
Since ellison is partly responsible, since he presented the affidavit to the court as evidence, he should probably be disbarred and removed from office.
And you can be totally sure that's not just partisan bias. It's not like the comment section indicates a fair percentage of the opposition party wrote a comment saying he's unethical and corrupt - without a single one of them naming an example.
The politicians are a symptom, the sickness is people who call them guilty or innocent based solely on the letter after their name.
Steve, you rock!
I would like to see that cross examination too. Thank you.
What an expert Dope!!!
I’m very curious how he billed this out, as you mentioned🤔
If you "make something up to prove a point," it means you are creating a fictional scenario or example, often exaggerated, to illustrate your argument and convince someone of your perspective, even if the specific details aren't entirely true; essentially, using a hypothetical situation to strengthen your point.
Key points about "making something up to prove a point":
• Not meant to be literal: The fabricated story or example is not intended to be taken as factual, but rather as a tool to explain a concept or idea more clearly.
• Can be risky: While useful in certain situations, using made-up stories can sometimes be seen as manipulative or dishonest if not used carefully.
• Important to clarify: If you are using a hypothetical example, it's crucial to explicitly state that it's fictional and not meant to represent a real situation.
Example:
• "Imagine a world where everyone only wore one color of clothing - this would clearly demonstrate how diversity can add richness to society."
Generative AI is experimental.
If the purpose of the expert witness was about the risks of misinformation by AI, wouldn't the AG actually want to keep that affidavit, but turn it into evidence of AI misinformation? It just seems to me this is a case of the worst possible action can be used for the best possible outcome.
Ben pinned by the guard of the sword on Steve's RHS
I'm blind today... good eye Bob.
@@BenLeitch It's an hour earlier than normal, maybe I'm not as sleepy...
Morning Ben
I missed it!
@@terrancecoard388 Don't worry Terrance, I think a lot of hunters will have missed this one.
@@Bobs-Wrigles5555 This was one of his sneakier placements. I am good at checking inside the handles...and on top of the microphone! Missing that one really hurt.
Tip of the iceberg that is the paid-expert industry.
Excluding punctuation, I proofread my own comments before posting. It's not 100% foolproof but it keeps me from posting something I would probably regret later.
I try that, but I read mine after I post, and often find I didn't write something I meant to write, or have typos to correct!
With all this talk about the Professor I am surprised Maryanne and Ginger were not mentioned.😅
I have a valid question. If you are accused of a crime by AI, do you still have the right to face your accuser?
Allow me to elaborate... your accuser, is, the court.
@@gregbillman42 I thought that in the US your accuser is "the people".
For a second, I was afraid the judge was going to be the AI culprit.
It’s also worth pointing out that he never had any authority to claim to be an expert in the first place. None of his qualifications relate to AI or AI safety. He is an internet archivist. It is outrageous that he was allowed into the court at all.
"I do not believe that word means what you think it means."
Inconceivable!!!!
What word "expert" or "ai". LOL
"Deep Fakes?" Like trying to convince a court of law that you've written something you haven't, and submitting an affidavit, including citations that don't exist?
So I wanna know how 'hallucination' aka telling lies got programmed into AI ?.
I've wondered the same thing. Is it being fed (programmed) with bad information to begin with...or is it taking a life of its own and crating alternative facts with said information. Next thing you know it will say you can divide by zero!
It didn't. The AI is familiar with how citations should exist in a certain type of document, and when it's given a prompt to recarate a document of that type, it's going to essentially copy that format with what it creates. Problem is an AI doesn't know what's true or what's not, it's just putting characters onto the page to make The document it generates look like it should. It doesn't know it's something that doesn't exist, because it didn't look up the citation. It just scraped the internet for information related to the prompt, and put text on a page in the right format.
Our current ai is not programmed. It's just a fancy autocomplete. It has no concept of facts, knowledge or anything. All it does is predict the new characters
AI is a misnomer it's machine learning. It reads a ton and paraphrases it back to you. So if it read false info it leaks into it's results. A lot of these models were just given access to opinion websites like reddit, so they have junk in junk out. That's why good says it's good to eat several small rocks a day, because it was a reddit comment, a joke. But the "AI" has no actual learning elements it just generates text. Real AI has to go through hundreds of thousands of iterations with punishment/reward parameters. Can't do that for these kinds of chatbots because you need a human to fact check every iterations. It's fancy auto complete and that is all it will ever be. It's all a ponzi scheme to milk money out of investors and the stock market.
@@NEPAAlchey Since I started putting glue on my pizzas, the cheese hasn't slid off /s
I'm Pro Se, I use gpt to write drafts, but I check all citations including the actual verbage in the case. Gpt still makes up case law all the time....
... so the old saying about people who represent themselves is true?
@@danwells9525 I've won my cases so far.
Are you too incompetent to represent yourself..?
That "expert" cannot testify as an expert ever again, but he still has tenure, so his 'career' is intact.
I wonder. Moral turpitude might be something he can be fired for.
Whether 'thrown out and resubmission not allowed' or 'career ruined'... Expert witnesses and attorneys (future judges) should consistenly suffer the consequences prescribed by law... else the importance of justice and of following the law appears as arbitrary (often as weighted in favor of the professional/privileged class and/or those in positions of wealth or power). This seems to be one of the things that has allowed (or is simply another example of) corruption to permeate every level of our society.
He’s a dumb professor, he can’t be fired!
Professor had his TA write it, which is why he "had no idea".
As a Minnesotan, the sad part is that there is nothing about this that surprises me. It's pretty much a case of " Stupidity As Normal." But, in Ellison's defence, he has to submit breifs with Artificial Intelligence in them. It's my understanding that there needs to be some sort of intelligence in court proceedings, and he doesn't have any Real Intelligence of his own.
Considering who hired him, gov timmy, no surprise at what we got. Now, at least the world knows how much of a clown walz is after a failed campaign.
FYI, if using a generative LLM (such as chatGPT) preface your question with "Do not make up responses." and the LLM will produce fewer fictious answers.
Ahahahaha. This is funny. That's not how LLMs work.
You may get better results if you include the citation documents as attachments (inline or RAG) to the prompt. But you still need to double check the result.
@@DVSProductions Try it both ways. By explicitly adding a prohibition against making up responses you are creating a "punishment" condition that doesn't exist by default.
@@brianorca Absolutely! I did say "fewer" fictitious results, lol.
*I think Perjury is a Crime! What is the penalty for Perjury? I will bet the "Expert" will not suffer the penalty for Perjury!*
*He should now be a Professor in the State Prison! 2024 Minnesota Statutes - 609.48 PERJURY. Just-Us!*
Doesn't perjury have to be intentional?
@@monkemode8128 Didn't he intentionally do that though?
If he would have had some other person write that for him and then passed it on as his own without checking it, would that be any different?
"No I did not kill that man, I only told paid that other man to do it."
@@justskip4595 IDK I'm not a lawyer. I would hope it's considered some kind of negligence though.
@ From my foreign point of view, there's insane amount of bizzare scapegoating happening in USA like it appears to be in this case too.
People who are obviously responsible being let go because someone lower down in the pecking order is put out as a scape goat to take the blame or something non human is to blame.
I do not have the words to express myself how insane this appears to me.
The expert's biggest mistake was failing to verify the citations. But, once they were found to be false, that called into question the supposed facts stated in the declaration. 😮
No. The biggest mistake was claiming to be an expert in AI and failing to understand the fact that it can't write your opinion
It also suggests that his opinions weren't his opinions.
You said what will be the downfall of the modern civilization. Critical thinking skills. It is not taught anymore. Children are taught what to think, not how to think. Now couple this whith a short attention span and the desire for immediate gratification.
The primary cause of being taught "what to think, not how to think" was the automation of education via machine-graded multiple choice tests, bring things back around to lazy, blind misuse of computers.
I stopped reading half way through. I can't tell if you make sense or not. Look, a squirrel!
@@ED-es2qv it is okay; there will not be a test.
It's called an After-David. Usually approved by a notary Republic, and signed under penance and perjury.
Truly, this one sounds like it could have been written by The Onion. Though, this (admittedly wonderfully ironic) example of hazards of using AI in a court case about the hazards of AI pales before The Onion’s use of parody in that legendary amicus brief. (:
Going through your back catalog as well as the current videos you put out, I miss the long videos.
When oh when will Keith Ellison stop making the news as one of the least competent people in politics?
Of COURSE they did.
This seems like performance art. 😂
Minor point: the software that the professor uses for academic papers is probably a citation manager which won’t have a baked in function to double check chatGPT’s work. They let you import full citations from a known good database (such as pubmed) and format them to the style desired by the journal you submit to. They can’t check an arbitrary exported list of citations.
However, it could be possible to have ChatGPT generate a list of unique identifiers for its citations (ie ISBN, PMID) and then manually check them either in the citation manager or something like Google scholar. That is more work than “trust me bro” obviously.
Microsofts AI understood why I don't like CO-PILOT and agreed that CO-PILOT is creepy....
That's another AI snafu that Leonard French made a video about, a couple of days ago. Since Co-Pilot is part of 365 (what they renamed Office), and lawyers and courts use Word documents, which is part of 365/Office. Co-Pilot risks the confidentiality of all those documents !
@@michaelmoorrees3585 I was using the one built into bing thing.... I asked it who was using copilot and actually liked it. The AI didn't really answer my question but it asked me if I plan on using copilot and I answered "No, I abore the idiea of copilot and it creeps me out".
The AI said it understood that copilot is obtrusive or something to that nature... and many people are not in favor of it.
Well, what is an expert?
X is an unknown, and a spurt is a drip under pressure.
Expert is therefore an unknown drip under pressure.
Expert= has been drip
You can easily train the Ai to use from the list of court cases or even use the look up tool it's self.
I assume he's tenured? If you mentioned it I missed it. But would that be enough to save his job?
I work for a religious organization this reminds me of an Ethical AI Use class.
"AI can be a powerful tool but it cannot and should not be what is delivered by messaging. Like any task you are the steward of the work and are responsible for everything written "
I just heard of this story the other day on Leonard French's channel
Ben is sandwiched between the sword hilt and bookcase to Steve’s right shoulder.
My dad worked as a technician in an oil refinery when I was growing up. One part of his job was sighing off on equipment that needed work. Usually such work was performed by outside contractors who may or (more usually) may not be failure with the operation of such equipment. There are a great many things you can do in an oil refinery that can make you and others unalive very quickly.
For this reason, when such work is needed a technician needs to make sure the equipment is safe to work on.
He never delegated this work to others. If he was signing off on it, he wanted to make sure it was, in fact, safe to work on. The consequences, go well beyond being responsible for killing people, if he signed off that something was safe when it wasn't could mean prison.
An accident took place in a separate refinery to a pipe that carried molten sulfur was empty, not under pressure and safe. It was not. The contractors opened the line and several tons of molten sulfur sprayed out and killed 15 people. Almost the entire staff of a particularly good contractor.
I believe the technician who signed off the equipment was safe spent the rest of his life in prison.
Little pro tip for anyone using AI to write anything. Do NOT copy-paste the output. Re-type the whole thing trying to rephrase things in the way that you normally would where it makes sense to do so. Some times some ideas are so simple that there is only one way to express it without overcomplicating it for the sake of saying it differently.
That is outside of the obvious things like following every single thing that it links to as a source and making triple sure that the reference exists, is correct, is reliable, and that the reference is being done correctly. I should say that I thought those were so obvious that they were not even worth mentioning, but clearly not....
The real pro tip is to not use AI for anything important.
Speaking of billable hours, does anyone remember the WPPSS debacle in the PNW? And how the lawyers were flying across country to bill more hours each day they worked on the lawsuits?
I'm an attorney in NY, and about once a year I catch another attorney using a block quote that omits words like "not" to alter the court's determination in their cite.
Its a demonstration of the peril first hand
Perjury is illegal at Bushwood Sir, And I NEVER lie!
I heard about this case. 🤣🤣🤣 The AG deserves this. "Are you saying we need to ask our experts if they use AI and how?" YES. That is exactly what we are saying.
I built my first computer, S100 bus, in the late 1970s. The older I get, the less I trust some of the things, programmers try to use computers for.
But the computer said....
Easy enough, the court should have it's clerk run all filings through AI to verify citations. I bet Chat GPT is pretty accurate at verifying citations as opposed to finding an important relevant citation in context of the legal theory you are pursuing.
He could be an expert in AI hallucinations?
Use of AI should be banned from court. Those who do should be held in comtempt and have their law license suspended for a month.
There might be an appropriate use for it, but with better safeguards than exist today.
@TheRealScooterGuy traffic lights 🚦
I'm tired of waiting for non-existent cross traffic. Cameras can look for cars and trigger the sequence if none are coming. Could also trigger all red when an emergency vehicle is coming or clear the intersection if the intersection is clogged.
@@tony_25or6to4 -- Agreed. I've said much the same elsewhere. But don't limit it to standalone management of one intersection. Let one intersection tell the next what is coming, and when. Let the system notice that construction has affected the traffic, and adjust accordingly. Teach it to recognize what happens with traffic when the weather changes. Also, give it multiple cameras and sensors watching overlapping sections of roadways, and recognize when one sensor or camera is sending false data, so that is can be scheduled for maintenance.
But whatever we do, don't let it write legal briefs until it can tell the difference between truth and falsehood in its own writing.
There should be a law that any lawyer filing false stuff is barred from any and all activities of a lawyer.
Steve's idea of a good time (in the legal sphere) is interesting. He wasn't kidding when he said he could go much, much longer.
Did you get a free bowl of soup for wearing that shirt? It looks good on you though.
Not only "How much do I have to pay you", but you'd have to pay more to move to the head of the line of lawyers wanting to cross examine him.
The more articles, legal cases, and lectures I see by college professors and so-called learned experts, I mourn for the educational system. With a few exceptions, most of them seem to be full of it.
The funny thing about all this is the AIs will straight up tell you that they hallucinate... I asked CoPilot about hallucinations and how to correct them. It said that it does not know how to correct itself, so usually it just fills in the blanks to make the item look more complete, and relies on people to review what it writes for accuracy. When I objected that that was dishonest and unethical, it said that it did not understand ethics.
I'm not sure that people do, either.
Hard to argue he isn't an expert in the danger of AI now. Might be the leading expert on this subject!
What is the "penalty for Perjury"? Is there a statute of limitations? has this penalty been imposed for this particular type of Perjury?
He’s more of an expert today than he was before this was discovered!
Please please please companies I am begging you to use AI in your cases, it provides so much more work for us lawyers in the long run!
Best use of a Simpsons reference in a titlecard EVER!
He should be done as a professor too. Academic dishonesty matters. How can he expect students to be different than him?
Wow. That's like going to court for a grand theft auto charge, with another stolen car.
You know something that scares me is I've been hearing recently on the news how they're trying to put AI into hospitals and healthcare and diagnosing stuff. It scares me because of all these mistakes that AI makes