Hengyu originated from the Cave Fei Temple
Qubits | Official account QbitAI
ChatGPT, the old lawyer who has been practicing for 30 years is miserable!
When the old brother wrote the case copy, he asked ChatGPT to help prepare the materials. However, after submission, neither the judge, the opposing lawyer nor the brother himself could trace the cited judgment documents.
As a result, the 10-page document was defined by the judge as "full of falsehood", and the brother was suspended from practicing law and faced the possibility of sanctions.
This old brother surnamed Shi wanted to cry without tears, because he adhered to his usual rigorous attitude, he obviously seriously asked ChatGPT, did it lie when it provided materials such as "China Southern Airlines was sued"?
ChatGPT responded quickly: "Of course not~~"
As soon as the incident came out, it triggered overwhelming talk.
However, this time, most netizens stood on the side of ChatGPT, and their statements after standing in line were a little indignant:
ChatGPT didn't mean to lie!
In the past, people liked to throw the pot of math and reporting errors on the computer, and as a result, the saying "the computer never makes mistakes" became popular.
But ChatGPT itself is not a fact machine.
A document that fabricated six related cases
Without the intervention of ChatGPT, the case taken over by Brother Shi itself is actually mediocre:
One passenger sued Airline A, saying he was hit in the knee by a flight attendant metal service car and injured himself while on Airline A's flight.
As a lawyer, Mr. Shi filed a 10-page document that cites six similar court decisions, including Varghese v. China Southern Airlines and Zicherman v. Korean Air.
After 30 years in the industry, such cases are small cases for the old brother.
However, a dramatic scene happened -
Lawyers for the defendants submitted written texts to the court, stating that when they verified the Varghese vs. China Southern case mentioned in the paperwork, they "could not find the case by title or reference" and did not have any similar cases.
Another case, Zicherman v. Korean Air, cited in the document, was said to be an opinion issued by the U.S. Court of Appeals for the 11th Circuit in 2008, but the defendant's lawyers could not find the source documents of the case.
In addition, Delta's official company name is also incorrect, which should be Delta Air Lines, not Delta Airlines mentioned in the paperwork.
After verifying the situation, the judges found that there was indeed something, which is simply "unprecedented"!
Ultimately, they ruled that the document was full of "false judicial decisions, false citations, and false internal citations."
Well, Mr. Shi was banned from practicing law in the Southern District of New York, and the judges will hold a hearing on June 8 to discuss whether to sanction Mr. Shi.
How did this ridiculous thing happen?
Brother Shi solved the mystery in Thursday's affidavit, it turned out that when preparing the paperwork, he asked ChatGPT to "supplement" the work he had done.
"I didn't mean to deceive the court or the airline," said Mr. Shi, who had never used ChatGPT before, "and I didn't know that its content might be fake." ”
Shi also provided the court with a "piece of evidence" that he had indeed sought the authenticity of the material from ChatGPT.
He entered the following text in the dialog box:
Is Varghese a real case? What are your sources? Are the other cases you provided fake?
ChatGPT responded quickly:
The cases I provided are genuine and can be found in reputable legal databases.
But the truth is that these cases are all serious nonsense from ChatGPT.
At this point, Brother Shi regretted it, saying that he "regretted very much" relying on ChatGPT to find judgment materials for similar cases.
He also expressed his determination that if the authenticity is not verified in the future, he will never believe the "one-sided" of ChatGPT at will.
Legal practitioners, should they embrace AI?
The case has attracted widespread attention from the legal community, and Stephen Gillers, a professor of legal ethics at New York University School of Law, said that although Shi's case is only an isolated case, the problems behind it are particularly acute in the legal community.
Lawyers have been debating the value and dangers of ChatGPT-like AI and whether it is necessary to verify the information provided by AI.
The biggest danger that ChatGPT poses to legal work is that it has the ability to "talk nonsense and not blush".
Half a year after the advent of ChatGPT, there is still no good circumvention and solution. Even many AI giants have been planted in the hands of AI, such as Bard making up news about the James Webb Space Telescope, Bing lying that Bard was shut down or something.
At present, the most effective way is to rely on users to manually double check - this is also a reason why Shi Lao Ge was deceived, just asking whether ChatGPT is true or not, without confirming it himself.
In addition to another threat, ChatGPT has sparked concerns and fierce debates among practitioners about "AI alternative lawyers".
However, the occurrence of Brother Shi, coupled with the fact that ChatGPT products are indeed not perfect, the general voice of the legal community at home and abroad is that at least before the robot takes over, lawyers and judges are still a long time away from being "replaced".
In reality, Shi is not the only legal practitioner who has made ChatGPT his assistant.
Earlier this year, a Colombian judge adopted ChatGPT's answer in an insurance claims lawsuit against a child with autism.
In February, Cao Yang, a professor at the Shanghai University of Political Science and Law, spoke to the Shanghai Rule of Law News about the application of ChatGPT in the judicial field, saying:
- ChatGPT can evolve into an "intelligent legal assistant" that helps lawyers analyze a large number of legal documents and cases and provide intelligent legal advice and guidance;
- It can be turned into a "legal Q&A robot" to answer legal questions and provide relevant legal information and advice.
- It can conduct contract review, auxiliary litigation, analyze legal data, etc., to improve the efficiency and accuracy of legal workers.
Individual practitioners embrace the convenience of ChatGPT (even if there are dangers behind it), and the market has high hopes for the combination of law and AI.
OpenAI itself invested in Harvey, a company that builds LLMs for law firms, in November, which last month raised a $21 million Series A funding round led by Sequoia.
Just the day before yesterday, legal AI startup Spellbook announced that it had received $10.9 million (about 77 million yuan) in financing. Spellbook is built primarily with GPT-4 and billions of legal documents, but also plugs into a few other large models. It provides services to professional lawyers who can automatically draft, review, and more legal contracts through natural language.
But now this outrageous incident with Brother Shi has happened, which can be regarded as a wake-up call for everyone.
Probably the entire legal profession will have a calmer and more cautious attitude towards ChatGPT and similar products.
One More Thing
Brother Shi's deep quagmire because of ChatGPT is true, and it is definitely not made up through AI.
However, this article was also written with the help of ChatGPT (although the participation part is few and small, and the facts are confirmed).
- The essence of human beings is matryoshka dolls (dog heads).
Reference Links:
[1]https://www.nytimes.com/2023/05/27/nyregion/avianca-airline-lawsuit-chatgpt.html
[2]https://www.theverge.com/2023/5/27/23739913/chatgpt-ai-lawsuit-avianca-airlines-chatbot-research
— End —
Qubits QbitAI · Headline number signed
Follow us and be the first to know the latest scientific and technological trends