Re:GPT Takes the Bar Exam
by corford on Monday January 09, 2023 @06:04AM
(#34309058)
As a counter to this, I found a recent Guardian article on ChatGPT did a nice job of cutting through some of the more optimistic hyperbole surrounding LLMs/AI at the moment and offered instead a more grounded perspective as to what ChatGPT is and what it is not: https://www.theguardian.com/commentisfree/2023/jan/07/chatgp...
[ Parent
| on HN ]
try it. get on an use it. It is pretty remarkable. I have been using for all sorts of stuff. It is a brain extension- the ways it is limited at the moment are based on what the builders intended "no, I won't let you build a virtual environment, no I won't roll up a dnd character etc.." but when pressed/jail broken it most certainly can do those things..
I feel like I felt when explaining google search in 1999
I have tried it and it's great. But, because it's so effective, it's easy to over extrapolate and anthropomorphise it. I think it's useful to keep in perspective (as the article and referenced paper - https://arxiv.org/abs/2212.03551 - talk about) that ChatGPT/LLMs are, ultimately, just extremely powerful statistical next-token prediction systems.
That doesn't at all invalidate the achievement or belittle the impact such systems will have on society but, keeping this in mind does help to avoid blowing everything out of proportion (which is easy to do because, yep, ChatGPT is genuinely very cool and exciting tech).