A lawyer, presumably in a hurry, submitted documents to the court that included references to “hallucinated cases.” Where did these hallucinations come from?
ChatGPT, of course. Judge Kevin Castel (S.D.N.Y.) is threatening the lawyer with sanctions.
ChatGPT has an annoying habit of making up information (hallucinating) rather than saying, “I have no idea.” And this lawyer didn’t realize that was the case, so he thought when ChatGPT said, “Varghese v China South Airlines Ltd, 925 F.3d 1339 (11th Cir. 2019)” was a relevant case, it was a real and relevant case.
It was not. It doesn’t exist.
Now, whether or not the court will go through with the sanctions, one thing is pretty straightforward: Lawyers aren’t going to lose their jobs to AI any time soon.
Yet, people panic about job loss due to AI–but not their jobs, someone else’s job.
A Pew study found that 62 percent of American workers believed that AI would have a major impact on workers in general, but only 28 percent are concerned that it will have a major effect on them personally.
That seems to be somewhat of a disconnect, but I suggest it’s not due to the naivete of the respondents but rather because no one really knows what other people do.
In December, Reuters ran an article titled “Will ChatGPT make lawyers obsolete? (Hint: be afraid).” It discussed Suffolk University Law School Dean Andrew Perlman, who wrote an article in a record-breaking time of one hour, using ChatGPT as a helper. Perlman concluded that ChatGPT wasn’t quite ready for prime time but that it wasn’t all that bad either.
But the problem comes not with expert use of ChatGPT but layperson usage. Perlman’s questions and the answers he received were pretty good and pretty accurate, but he knows that because he’s an expert. If you’re not an expert, you cannot know what is correct.
Why do people think other people’s jobs will be affected
Many years ago, my job share partner and I ran layoffs for a Fortune 100 company. The company wanted to outsource all HR administrative tasks, and our jobs got put on the chopping block. We said, “This isn’t a good idea,” but the decision-makers insisted that it was.
Shortly before the outsourcing began, the company requested a list of our tasks. So we wrote one. I don’t remember how long the document was, but it included writing legal documents, coaching managers, serving as a subject matter expert in all terminations, ensuring legal compliance, etc. The outsourcers responded that they could not take on this work.
What had happened?
Whoever chose to outsource our positions only saw the end product: Folders filled with legal documents personalized to each employee, which explained the benefits the former employees would receive. They assumed our job was printing and stuffing these documents in folders. They literally had no idea that we did anything other than hit print.
Often we see other’s jobs like this–just the end product. So, of course, it can be automated. Anyone can hit print!
But you know what happens in your own day. You know that AI can’t easily replace what you do.
But it will have an impact.
How will AI impact your job?
“Goldman Sachs says generative A.I. could impact 300 million jobs — here’s which ones,” says a headline, and then the article goes on to state:
“In the U.S., office and administrative support jobs have the highest proportion of tasks that could be automated with 46%, followed by 44% for legal work and 37% for tasks within architecture and engineering.”
The reality is AI has already affected your job. AI is nothing more than a fancy algorithm that pulls information together. ChatGPT uses natural language, so it sounds like you’re talking to a human, but algorithms returning information aren’t new. If you’re old enough, you remember when you used to have to go to the library to find out information on Tanzania rather than doing a 5 second Google search.
Way back in 2019, AI was already in use in numerous areas, including (according to Vikram Singh Bisen)
Virtual Assistants or Chatbots
AI has impacted your job. And it will affect your career further. The key is that you need to learn how to use it. The unfortunate lawyer above thought he could use it to save time, but instead may end up with severe punishment.
Keep in mind the following things:
ChatGPT is like a know-it-all. It hates to admit that it doesn’t know something, so it makes things up.
It is not creative. Nothing new comes out of ChatGPT; it can only recycle older ideas. It seems new to you because you haven’t read the whole internet, but the ideas are old.
ChatGPT is always a starting point–not an endpoint. So use it to create rough–but not final–drafts.
Security is still an issue. Be careful not to give ChatGPT any confidential information.
You can use great prompts to get help with whatever problems you come up with, but only subject matter experts can know whether to trust the information or not. And even then, if it tells you anything you didn’t already know, you’ll need to double-check–probably using another algorithm called Google.