Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

1-800-ChatGPT

help.openai.com

Post reply on HN
Print view View original

341–350 of 350 posts

Re: 1-800-ChatGPT

#341

Post by whimsicalism »

semi-extrinsic wrote: post #136
whimsicalism wrote: post #97

Earlier quoted context omitted.

lots of coding tasks, discussions about physics/QM. I find that it produces better quality answers than 4o, which often will have subtle but simple mistakes. Even writing, where it is supposed to be worse than 4O, I feel that is does better/has a more solid understanding of provided documents.

> discussions about physics/QM Interesting, could you share an example of this where it provides something of value? I've tried asking a few different LLMs to explain renormalization group theory, and it always goes off the rails in five questions or less.

sorry missed your reply. I realized most of the stuff I was asking about QM was actually to 4o, it’s mostly stuff about things covered in this book https://en.wikipedia.org/wiki/Quantum_Field_Theory_in_a_Nuts...

Re: 1-800-ChatGPT

#342

Post by lethibo »

lxgr wrote: post #169
hypeatei wrote: post #49

Earlier quoted context omitted.

I agree about an abundance of apps, but what type of value are LLMs adding? It can sometimes be useful to input a more "human" search and have something get spit out but 60% of the time it completely lies to you. I'm talking about questions related to web specifications which are public documents. Section numbers, standards names, etc.. will be completely made up.

That's because we're currently largely not using them correctly, i.e. hooked up to RAG instead of hoping that they've memorized enough of the training data verbatim, which is arguably a waste of neurons in a foundational model. Imaging being graded on your ability to quote exact line numbers of particular parts of your codebase as a senior software engineer without being able to look at it! LLMs are not, in isolation…

[dead]

Re: 1-800-ChatGPT

#343

Post by crazygringo »

jazzyjackson wrote: post #300
aoeusnth1 wrote: post #293

Earlier quoted context omitted.

OP: good movies are better than bad movies. Replies: you buffoon, you actual clown. How dare you value good things more than bad things. I can’t take this knee jerk response seriously. Why wouldn’t good movies be more worthwhile than bad movies? How is this even controversial?

Because whether a movie is good or not is not an objective, one dimensional thing that can be represented by a score on rotten tomatoes?

It's not perfect, but it's a very strong and useful signal.

Never in my life have I seen something with 98% and thought, well that was a crap movie.

And never in my life have I seen something with 35% and thought, that was amazing!

It's more in the 75-90% range where you have to consider the "dimensionality" of the thing, like whether it's a genre you like, or which individual reviewers match your tastes more precisely.

Re: 1-800-ChatGPT

#344

Post by jazzyjackson »

crazygringo wrote: post #343
jazzyjackson wrote: post #300

Earlier quoted context omitted.

Because whether a movie is good or not is not an objective, one dimensional thing that can be represented by a score on rotten tomatoes?

It's not perfect, but it's a very strong and useful signal. Never in my life have I seen something with 98% and thought, well that was a crap movie. And never in my life have I seen something with 35% and thought, that was amazing! It's more in the 75-90% range where you have to consider the "dimensionality" of the thing, like whether it's a genre you like, or which individual reviewers match your tastes more precise…

Yeah I guess you're right, even the "so bad its good" movies get points for being campy and the love it / hate it reviews average out.

Re: 1-800-ChatGPT

#345

Post by mostertoaster »

crazygringo wrote: post #256
frakt0x90 wrote: post #215

Earlier quoted context omitted.

I get it's just an example, but are we really this far gone? Just watch a random movie and if you don't like it, pick another. This is such an extreme micro-optimization of a small experience.

You don't often know if a movie is good until you've finished it, because it all depends on how the story came together in the end. You can spend 2 hours watching a moving, emotional story that teaches you something new about the human condition and the choices we make in our lives. Or you can spend 2 hours that turns out to be full of plot holes and inconsistent characters, where nothing makes sense in the end and y…

The valuable time we save in not watching bad movies gives us more time to argue on hacker news, so a good optimization.

Re: 1-800-ChatGPT

#346

Post by dartos »

kbelder wrote: post #340
glenstein wrote: post #246

Earlier quoted context omitted.

You're still trying to imply that the list as a whole is as inaccurate as that one particular example. And I think that's quite obviously not the case, most, probably every other example on the list is just fine.

For comparisons, Wikipedia's opening paragraph for Anora reads: "Anora is a 2024 American comedy-drama film written, directed, and edited by Sean Baker. It follows the beleaguered marriage between Anora (Mikey Madison), a young sex worker, and Vanya Zakharov (Mark Eydelshteyn), the son of a Russian oligarch. The supporting cast includes Yura Borisov, Karren Karagulian, Vache Tovmasyan, and Aleksei Serebryakov." I hav…

You see the issue here, right?

Instead of just checking with a first party source, you ask a statistical guessing machine for an answer.

There was a disagreement about the answer, so we needed to dig deeper.

You bring up Wikipedia, a 3rd party source of information. That description could also be wrong (it’s probably not, but stick with me)

Instead of just checking with a first party source (IMDb is very easy to search on), we went through several layers of obfuscation.

This was an issue for Wikipedia early on, but it has citations, at least. AI doesn’t and doesn’t have an army of people constantly fact checking every answer generated either.

There’s no benefit to asking AI for information like this. Especially since the in flight summary has accurate information that’s more than “drama, sex worker, cony island”

Maybe something like perplexity is better, since it has citations, but I haven’t tried it for very long yet.

Re: 1-800-ChatGPT

#347

Post by qwertox »

cj wrote: post #238
paxys wrote: post #83

Earlier quoted context omitted.

Regular human conversational voice, especially over the phone, is going to be a gold mine for training customer support AI agents. Actors reading movie scripts can't really provide that amount of relevance.

“This call is being recorded for quality and training purposes” truly has a new meaning.

Wow... I never thought of this.

Re: 1-800-ChatGPT

#348

Post by crazygringo »

d0gsg0w00f wrote: post #307
crazygringo wrote: post #256

Earlier quoted context omitted.

You don't often know if a movie is good until you've finished it, because it all depends on how the story came together in the end. You can spend 2 hours watching a moving, emotional story that teaches you something new about the human condition and the choices we make in our lives. Or you can spend 2 hours that turns out to be full of plot holes and inconsistent characters, where nothing makes sense in the end and y…

Because life just isn't perfect. If you go seeking this perfect optimization in every aspect of your life then the world is going to be a continual disappointment.

Who said anything about perfection? About life being perfect, or about perfect optimization? Or about expecting perfection?

You can just pick a better movie and make life better.

What is your objection to that?

Re: 1-800-ChatGPT

#349

Post by crazygringo »

zeroonetwothree wrote: post #287
crazygringo wrote: post #256

Earlier quoted context omitted.

You don't often know if a movie is good until you've finished it, because it all depends on how the story came together in the end. You can spend 2 hours watching a moving, emotional story that teaches you something new about the human condition and the choices we make in our lives. Or you can spend 2 hours that turns out to be full of plot holes and inconsistent characters, where nothing makes sense in the end and y…

If you aren’t enjoying it then stop watching. And if you did enjoy it for 2 hours who cares what the review says.

I think you missed the very first sentence of my comment.

Re: 1-800-ChatGPT

#350

Post by etherealG »

frakt0x90 wrote: post #215
bsimpson wrote: post #168

On coast to coast flights, there's often not a good way of knowing what movies are available until after you've left cell coverage. This makes simple research like checking the IMDb score challenging. Alaska Air has a whitelist of messaging services that you can use for free during the flight. WhatsApp is on that list. So if you want to research obscure plane movies on an Alaska flight, you can connect to their wifi…

I get it's just an example, but are we really this far gone? Just watch a random movie and if you don't like it, pick another. This is such an extreme micro-optimization of a small experience.

Why do you try to mandate how others experience the world?
Post reply on HN