LegalEagle: I'm Begging You, Stop Hiring AI Lawyers [video]
youtube.com
youtube.com
Whether or not AI lawyers could hypothetically be useful in 10 years, today's LLMs are observably and obviously not appropriate. A legal AI needs to have a concrete idea of sources it is citing. It needs to have some sort of equivalent of a human pulling down a book and knowing (however you choose to determine the meaning of that world) exactly what the reference refers to. Modern LLMs merely create "plausible sounding" citations. Which we've already seen with people using these technologies in the legal world.
But LLMs are not the be-all, end-all of "AI". Future AIs with different architectures and capabilities may be able to function as lawyers reasonably.
LLMs certainly can not today, though, just by the nature of their architecture.
I see this frequently when generating source code; it'll get me 80-90% there, but I need to fix, massage, and adjust. This is fairly easy, since the code-test-code loop is already part of a normal development cycle.
When it's a matter of law, though? You don't really get those kinds of do-overs, and if the judge catches you citing a hallucination in your argument? I doubt that would go over well.
I'm not sure the "CALL (833) 3-MY-BIRD" guy is an unbiased guide as to the value of AI legal stuff.
And it wasn't particularly down to bad lawyers - I phoned quite a few and the majority didn't even want to handle the work as they were busy and it's not big money relative to other stuff.