AI, so far, is really terrible at legal research. So bad it gets young lawyers who use it sanctioned on a regular basis.
Reposted from
Michael Caley
this is a good way of getting at the disconnect between how AI is advertised and how AI is used
AI is really good at stuff like coding, legal or medical research, visual pattern recognition, that allows reasonably trained users to expand their productivity
but the ads are about way dumber stuff
AI is really good at stuff like coding, legal or medical research, visual pattern recognition, that allows reasonably trained users to expand their productivity
but the ads are about way dumber stuff
Comments
I used to it write cover letters last year, it was easier to edit them than to write them myself, but geez they were terrible.
It might come up with better answers than Trump and Vance though.
Robots writing letters for robots to read.
I'd take code from an overseas underpaid sweatshop before I integrated AI generated code into production and that's not something I say lightly.
But I'm not a young lawyer.
I mean, the more AI gets woven into 'general internet', it's going to be harder and harder to stop it.
https://www.reuters.com/legal/new-york-lawyers-sanctioned-using-fake-chatgpt-cases-legal-brief-2023-06-22/
It's literally not good at any of those things, granted if this person meant like specifically using machine learning to detect breast cancer... yes it is good at that specifically because it's trained on that one thing.
As it still returns cases that don't exist, it's certainly not ready for legal research.
When real AI gets here it’s gonna be pissed.
*or animals
And this is how the AI scam propagates.
For comparison, think about how AI can't write a book: it can write a sentence, or a paragraph, but not anything that requires critical thinking or ingenuity.