220
Judge rebukes law firm using ChatGPT to justify $113,484.62 fee as “utterly and unusually unpersuasive”
(www.theregister.com)
this post was submitted on 25 Feb 2024
220 points (99.5% liked)
Technology
59402 readers
3048 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I've asked ChatGPT to explain maths to me before. I can't remember what it was but it was something when I knew the answer and was trying to calculate the starting value.
It told me the answer and I asked for the explanation. It went something like this (not actual, just a tribute):
Me: Uh, the answer is supposed to be 9,000,000.
ChatGPT: Sorry, it seems you are right. Here's the corrected version:
nods knowingly as if he doesn't suck ass at math and understands the mistake
Sometimes GPT says it’s using the correct values, but somehow gets the wrong answer regardless. Also the opposite happens frequently, and that’s when I realized I was pushing it too hard.
Don’t ask it to calculate the ratio between the surface areas of the Moon and Earth. Instead, ask what are the relevant radiuses are and calculate everything yourself.
So it did it correctly but you told it to hallucinate? Or did it just fail from the get to?
It really isn't great at math, but I've had okay results for equations where common integrals/trigonometry is used. It's quite easy to spot the mistakes and can lead you to the answer even if it's wrong in the explanation. Pretty much like how it can hallucinate while programming but still end up useful.
WolframAlpha is still my go to though if I'm lazy. But I haven't payed for it in ages.
It was a question like where I knew the answer and needed the correct value to put in the equation to get that end result. I wish there was a better way to search previous chats because it would help if I could remember the context. But anyway, the first time it got the maths right from the starting value to the ending value, but it didn't actually answer the question because the ending value was not the one I asked for. It was as good as if it gave me a random answer.
I pointed out that it hadn't answered the question, and that's when it just changed the last step to make it the answer I was looking for. It was supposed to adjust the starting value to make it have the correct outcome.
Haha yes, eager to please.
Yeah, I think what you needed was something of an inverse problem solution, for some calculations it may be quite hard to compute. Were you successful in finding the starting value without ChatGPT's help?
I can't remember! It was probably a year ago, but knowing me I probably tried random starting numbers until I got the answer I wanted.