Comment by cortesoft
2 days ago
I have some lawyer friends, who work as internal council to companies, that are already experiencing this.
People are cranking out legal requests and claims with LLMs and sending them to companies. Almost all of them are pretty much meaningless, and should be ignored.
However, they legally can't just ignore them. They have to have someone review the claim, verify that it is bullshit, and then they can ignore it. That takes time, though.
So people can generate and send millions of legal claim instantly, but the lawyers have to read them one by one.
The asymmetry of effort is huge, and causes real issues.
They just need to implement a better LLM that is able to deal with all that crap.
At the end we will just have agents interacting with each other.
The problem is the legal liability is also asymmetrical. If the person making the bogus legal claims with the LLM messes up, the claim will just be dismissed and it isn't a big deal for the person making the claim. If the reading LLM messes up and classifies a real claim with the bogus ones, the company could have a major problem.