Good AI hunting
When legal gibberish makes it to court
Earlier this year, “60 Minutes” asked Google AI program Bard to explain inflation. Bard shot back a report in seconds, complete with a suggested reading list. The books on the list were wholly fabricated out of thin air.
At that point, the writing was on the wall. Problems of this nature would surely crop up in other areas of life. Sure enough, they’ve reached the courtroom.
According to the Associated Press, last week two attorneys from the law firm Levidow, Levidow & Oberman in New York City “filed a lawsuit against an airline that included references to past court cases …” You guessed it, the cases were made up.
The tech industry has warned that one of the drawbacks of AI is that it tends to “hallucinate.” And hallucinate it did.
One of the precedents was “initially described as a wrongful-death claim brought by a woman against an airline only to morph into a legal claim about a man who missed a flight to New York and was forced to pay additional expenses.”
The judge asked the humiliated lawyer after the mishap was discovered: “Can we agree that’s legal gibberish?” Most first-year law school students could.
One can imagine the damage to the firm’s reputation. It would also be interesting to know how the report was billed to the client. What would’ve taken substantial hours by a human being took only a few seconds to prompt the program. We can’t wait to see how this is handled by the New York Bar.
There but for the grace of God . . . If this leaks into the legal profession, it’s probably already in the journalism trade. Get ready for it. And hope it doesn’t do any more damage to newspapers.
We should note that tech types have come out begging for regulation. Some say it could lead to human extinction. Those on less caffeine say AI could throw a lot of people out of work.
That’s scary enough, but there’s another reason the brakes should be applied to the AI monster truck, bringing it to a screeching halt: It’s not ready for prime time.
The tendency to hallucinate speaks of an individual who is of unsound mind, and we don’t want unsound minds, artificial or not, being responsible for things like supposed medical breakthroughs, the nuclear codes, or poetry.
Would you want somebody (or something) piloting your plane if there’s a chance your pilot may try to refuel at a landing strip in the mythical town of St. Artexomaha?
It also seems that if this technology is smart enough to mine the Internet at lightning speed for a legal document, somewhere along the way it would have learned that fabricating legal precedent is potentially grounds for disbarment.
To hijack a concept from the movie “Good Will Hunting,” AI may be able to write a report about everything anyone would ever want to know about Michelangelo and his famous artwork on the ceiling of the Sistine Chapel, but it will never know what it smells like inside the Sistine Chapel.
If AI is to work for the good of society, human beings must be there to raise it like a child who has more smarts than sense.