Generative AI is both one of the biggest trends in tech today, as well as among the biggest bogeymen in conversations of job security and our imagined welcoming of our robot overlords. While we gladly concede the former, the latter remains as farfetched today as when the conversation was still mostly about machine learning in the last decade. Recently, a lawyer in the US has just added new evidence to that, with a hilariously disastrous use ChatGPT in writing a legal brief.
As The New York Times reports, one Steven A. Schwartz, of law firm Levidow, Levidow & Oberman, prepared a court filing with the help of ChatGPT. The case itself involves one Roberto Mata, who is suing Avianca, Colombia’s biggest airline, because he claims to have been injured by a metal serving cart. The airline asked the court to throw the case out, prompting the legal brief in question to be written, arguing the case should proceed with citations to a number of similar cases and court decisions.
Unfortunately for Schwartz, who claims to be unfamiliar with ChatGPT, the cases mentioned in the brief did not actually exist as they were invented by the generative AI tech. The lawyer also claims to have asked the AI to verify that the cases were in fact real, to which it apparently said yes. He himself may face potential sanctions for his troubles for providing the court with “bogus judicial decisions, with bogus quotes and bogus internal citations”.
To cut the lawyer of 30 years some slack, we can see that this is a genuine mistake that someone could make. After all, on the surface, generative AI does seem impressive enough to warrant all the Terminator and Skynet memes. But all it takes is to go slightly beyond the surface to see that more often than not, the AI is winging it while trying to maintain grammatical coherence. Therefore, while AI may indeed take over the world in all the stereotypical ways one day, that day is easily decades away still.
(Source: NY Times)