There’s an ever-increasing ooze of news about AI in legal proceedings. Most of it seems to be true, like the noises when you’re stumbling around in the dark.
The AI is effectively being misused, and it’s not exactly helping the law do its job. A case in Australia is illustrative. The case was written by ChatGPT and cited 17 other cases which never existed as precedents. Not much help to the client.
That’s not the whole story. Fake citations seem to be a snowballing thing worldwide. They’re sometimes called “AI hallucinations” which is meaningless and dangerously euphemistic. The fundamental processes are all too obviously flawed. None of this should be happening at all.
For “due process”, you need “due diligence”. These cases as written up by the AI were effectively worse than useless. They are actual liabilities to the lawyers presenting them. In a suffice egregious case, they could get disbarred from practice.
Let’s take the idea of garbage as documentation out of the purely legal context:
You could get sued for a lot of things simply because there’s no adequate oversight on AI-provided information.
Imagine:
Buying a property and the owner’s name, if incorrectly AI-generated, is someone else.
Contracts made out to the wrong people.
Finance sector “hallucinations” could cost billions.
Accounts issues.
Basic inventory on any dashboard.
Identity chaos.
This could become a very long list, but you can see what an avalanche of train wrecks this could cause. AI service suppliers could well be held liable.
I’m getting tired of using the word “naivete” in context with AI. It’s getting monotonous.
Let’s use “unbelievable stupidity” instead.
There’s an unequivocal bottom line emerging here:
All AI work clearly needs to be oversighted by skilled people who know how to oversight.
Nothing AI generates can simply be taken on trust.
When it’s money or property involved, you can expect to get very thoroughly milked for screwups.
You’ve got absolutely no defense when it comes to errors in commercial transactions.
A class action with any and every line of AI script on every dud operation isn’t impossible.
Do you want a pack of Artificial Nutcases roaming the world creating problems at their processing speeds?
Meanwhile, there are serious technical issues:
Why are these ongoing serial AI idiocies being tolerated at all?
The flaws need to be identifiable and visibly traceable in their processes.
What’s wrong with the LLMs that such mistakes are possible?
The vexed issue of AI sabotage, by AI and people, needs to be sorted out.
For all those altruistic human-and-employee-hating mofos, get the message.
This cluster MUST be fixed, ASAP. It’s too dangerous to be allowed to continue.
________________________________________________________
Disclaimer
The opinions expressed in this Op-Ed are those of the author. They do not purport to reflect the opinions or views of the Digital Journal or its members.
