Ken White
ThenIWhispered
Dirk Hohndel@Popehat
Maybe you should ask the Intern instead of using ChatGPT?
Or were you thinking of replacing all interns by ChatGPT?
m01r4
Steve's Place@Popehat But, as training for future politicians avoiding answering questions from pesky reporters, not bad. Not bad at all.
Abstract Tesseract@Popehat How did people even answer these questions before large language models, I am amaze
Shinydan@Popehat is there extant case law on this matter?
sam
Rowyn@Shinydan Yup! @Popehat did a podcast episode about it: https://legaltalknetwork.com/podcasts/make-no-law/2018/11/the-f-bomb/
✨pencilears✨@Shinydan @Popehat Cohen vs California.
if this was an essay question on a test? ChatGPT just got it wrong.
https://wikipedia.org/wiki/Cohen_v._California
Bernie Luckily Does It
Steve Bellovin@Popehat Just don’t ask it for case citations to back up its answer…
Charlie Galvin@SteveBellovin Actually do ask it. Feed it a brief and ask it to Shepardize.
Stu Duerson, machina exo deusSooo...'Oliver Wendell is a Holmie' is out?
Darryl Ramm@Popehat Is it OK to laugh out loud when lawyers use a LLM/AI to set fire to themselves in a crowded courthouse?
Ulrike
TorontoWill@Popehat Initially my impression was that it was a good tool for search engines to use to synthesize answers to questions, but as I understand it better, answering questions is totally the wrong use of the tech. It doesn’t have a grasp of what words mean, just the patterns of how people use them. It’s always bullshitting, even when it’s right.
@TorontoWill @Popehat
"It's always bullshitting, even when it's right," may be the most concise description of LLM I've see thus far!
"It's always bullshitting, even when it's right," may be the most concise description of LLM I've see thus far!
Noah Cook@Popehat When I asked it if it was legal to wear a black armband in school, it at least cited Tinker.
An answer about the color of margarine surprisingly suggested that there were legal requirements in some jurisdictions.
But I think this is an issue of reliability, does it always return that answer? And I suspect it matters how you ask the question, mine might have been too specific, leading it to the correct answer.
Waterluvian@Popehat what about “ask the judge about jury nullification” ?
Trading Places Research@Popehat Maybe start using it to write briefs? https://simonwillison.net/2023/May/27/lawyer-chatgpt/
Thomas Kitchens@Popehat It still outperforms most people in that it entirely failed to mention yelling about fires in movie theaters.
Eric R. Smith@Popehat To be fair (to be faaiiir...) the question didn't specify a country. That particular attire is legal in the US, but I would be hesitant to try it here in Canada, and there are places in the world where it would definitely earn you an all expenses paid vacation in a less than desirable locale.
Dave@Popehat “One man’s vulgarity is another’s LLM prompt.”
Michael Olsen@Popehat sheesh, you'll be out of a job in no time 😜
Stephen Kiskamp@Popehat Do you think ChatGPT will get more/less accurate if podcasts are added to the training set? 🤔
(And, yes, I mean specifically this MakeNoLaw episode, but also realize that there are less accurate podcasts out there.)
Gojira1000@Popehat The bit that disturbs me is seeing so many people using the shitty thing at work/professionally