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

GPT-5 is behind schedule

wsj.com

Post reply on HN
Print view View original

771–780 of 1001 posts

Re: GPT-5 is behind schedule

#771

Post by radlad »

FrustratedMonky wrote: post #703
ithadtobe119 wrote: post #679

Earlier quoted context omitted.

I think I recall that being somewhat disputed because the relationship status was self reported, some suggested that men might not consider certain types of relationships as serious but women do, so there's a disparity in reporting what is and isn't an actual relationship and the reality might be more balanced. Sweden statistics, xd. From what I can find after a brief search, there's this one [0] that claims 63% for…

So are you saying, some gen Z men are in a relationship, but don't know it? I do buy that, it seems to be the basis of some rom-com plots. The clueless guy that doesn't know he's being reeled in. Other factor. As the other post suggested. There are large age gaps. Women date older, men date younger. This is also long known. Does it add up to 60/30? That does seem high, but maybe with every other factor thrown in, it…

> So are you saying, some gen Z men are in a relationship, but don't know it?

Or, you know, are leading women on.

Re: GPT-5 is behind schedule

#772

Post by jaybna »

Workaccount2 wrote: post #765
jaybna wrote: post #741

25% of the top 1000 websites are blocking OpenAI from crawling: https://originality.ai/ai-bot-blocking I am betting hundreds of thousands, rising to millions more little sites, will start blocking/gating this year. AI companies might license from big sources (you can see the blocking percentage went down), but they will be missing the long tail, where a lot of great novel training data lives. And then the big sites w…

Doing basic copyright analyses on model outputs is all that is needed. Check if the output contains copyright, block it if it does. Transformers aren't zettabyte sized archives with a smart searching algo, running around the web stuffing everything they can into their datacenter sized storage. They are typically a few dozen GB in size, if that. They don't copy data, they move vectors in a high dimensional space based…

So then copyrighted content scraped is not needed for training? Guess I missed AGI suddenly appearing that reasoned things out all by itself.

Re: GPT-5 is behind schedule

#773

Post by IAmGraydon »

mrbungie wrote: post #490
IAmGraydon wrote: post #466

Earlier quoted context omitted.

Making a pong clone by telling the LLM to make a pong clone is a cute trick that sometimes works, but that's not the way anyone who understands how to properly use these tools is using them. You don't describe and app and hope the LLM builds it correctly. You have to know how to architect an application and you use the LLM to build small pieces of code. For example, you tell it to build a function that does x, takes…

No true scottsman fallacy. I know how to use them, but using them "correctly" still produces many errors. They suck at non-trivial code outside of standard library usage and boilerplate coding: I gave an example and parent did as well. In that regard would at least change your phrase from "actual coders" to "actual senior coders", as any junior receiving bad advice (in eternal loops as LLMs normally like to do it) is…

My point is that while you do have to give them coding problems that would have appeared in their training set (I guess you could call that trivial), every coding problem becomes trivial when you break it down to it's constituent parts. As you know, the biggest applications are just a lot of very simple building blocks working together. The point of using LLMs to code is not to solve complex problems. It's just to write code you could have written yourself at the speed of light using a natural language interface.

The way you described using LLMs to code seems like the approach someone who doesn't know how to build software might take, which is why I used the wording I did. From that angle, I agree with you - I can't even get Sonnet to create a working prototype of a basic game from a prompt. That said, I'm using it to build a far more complex enterprise web app step by step by using it in the way I mentioned above. It does work for these things, but you have to already know how to do what the LLM is doing.

Re: GPT-5 is behind schedule

#774

Post by IAmGraydon »

margalabargala wrote: post #502
IAmGraydon wrote: post #466

Earlier quoted context omitted.

Making a pong clone by telling the LLM to make a pong clone is a cute trick that sometimes works, but that's not the way anyone who understands how to properly use these tools is using them. You don't describe and app and hope the LLM builds it correctly. You have to know how to architect an application and you use the LLM to build small pieces of code. For example, you tell it to build a function that does x, takes…

As a coder with some noncoder friends who have made some very impressive things with chatGPT, you're selling it short. It does both. It gives coders superpowers, and gives noncoders the ability to do things that would have previously taken them months, or another person.

Do you mind sharing what they've created with it?

Re: GPT-5 is behind schedule

#775

Post by radlad »

cowsaymoo wrote: post #624
saagarjha wrote: post #614

Earlier quoted context omitted.

Yes, but how do you know which is which?

That is also a broader epistemological question one could ask about truth on the internet or even truth in general. You have to interrogate reality

That's certainly true, but I think it's also true that you have more contextual information about the trustworthiness of what you're reading when you pick up a book, magazine, or load a website.

As a simple example, LLMs will happily incorporate "facts" learned from marketing material into it's knowledgebase and then regurgitate it as part of a summary on the topic.

Re: GPT-5 is behind schedule

#776

Post by Kostchei »

jaybna wrote: post #741

25% of the top 1000 websites are blocking OpenAI from crawling: https://originality.ai/ai-bot-blocking I am betting hundreds of thousands, rising to millions more little sites, will start blocking/gating this year. AI companies might license from big sources (you can see the blocking percentage went down), but they will be missing the long tail, where a lot of great novel training data lives. And then the big sites w…

Using the real world- as in vision, 3d orientation, physical sensors and building training regimes that augment the language models to be multidimensional and check that perception, that is the next step.

And there is very little shortage of data and experience in the actual world, as opposed to just the text internet. Can the current AI companies pivot to that? Or do you need to be worldlabs, or v2 of worldlabs?

Re: GPT-5 is behind schedule

#777

Post by shanusmagnus »

Kostchei wrote: post #776
jaybna wrote: post #741

25% of the top 1000 websites are blocking OpenAI from crawling: https://originality.ai/ai-bot-blocking I am betting hundreds of thousands, rising to millions more little sites, will start blocking/gating this year. AI companies might license from big sources (you can see the blocking percentage went down), but they will be missing the long tail, where a lot of great novel training data lives. And then the big sites w…

Using the real world- as in vision, 3d orientation, physical sensors and building training regimes that augment the language models to be multidimensional and check that perception, that is the next step. And there is very little shortage of data and experience in the actual world, as opposed to just the text internet. Can the current AI companies pivot to that? Or do you need to be worldlabs, or v2 of worldlabs?

Ironically, if it plays out this way, it will be the biggest boon to actual AGI development there could be -- the intelligence via text tokenization will be a limiting factor otherwise, imo.

Re: GPT-5 is behind schedule

#778

Post by knowaveragejoe »

reissbaker wrote: post #625
sollewitt wrote: post #493

Earlier quoted context omitted.

LLMs suffer from the "Igon Value Problem" https://rationalwiki.org/wiki/Igon_Value_Problem Similar to reading a pop sci book, you're getting an entertainment from a thing with no actual understanding of the source material rather than an education.

Earlier in this thread, people mention the counterpoint to this: they Google the information from the LLM and do more reading. It's an excellent starting point for researching a topic: you can't trust everything it says, but if you don't know where to start, it will very likely get you to a good place to start researching. Similarly, while you can't fully trust everything a journalist says, it's obviously better to h…

LLMs are pretty good at attacking the "you don't know what you don't know" problem on a given topic.

Re: GPT-5 is behind schedule

#779

Post by Workaccount2 »

singpolyma3 wrote: post #405
Workaccount2 wrote: post #359

Earlier quoted context omitted.

The buzzkill when you fire up the latest most powerful model only for it to tell you that peanut is not typically found in peanut butter and jelly sandwiches.

I don't think providing accurate answers to context free questions is even something anyone is seriously working on making them do. Using them that way is just a wrong use case.

People are working -very- seriously on trying to kill hallucinations. I'm not sure how you surmised the use case here, as nothing was given other than an example of a hallucination.

Re: GPT-5 is behind schedule

#780

Post by knowaveragejoe »

pino82 wrote: post #735
cbau wrote: post #718

Earlier quoted context omitted.

You can turn to actual experts, e.g. YouTube or books. But yes, I have recently had the misfortune of working with a personal trainer who was using ChatGPT to come up with training programs, and it felt confusing and like I was wasting time and money.

When I'm looking for actual experts, the first thing that comes to my mind is definitely YouTube!! And least when it's about YouTube specific topics, like where the like button and the subscribe button is. They will tell me. Every. Single. F*cking. 5. Minute. Clip. Again. And. Again. Not soooo much for anything actually important or interesting, though.... ;) PS: Also which of the always same ~5 shady companies their…

Unironically, youtube is a great place to find actual experts on a given subject.
Post reply on HN