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

GPT-5

openai.com

Post reply on HN
Print view View original

751–760 of 1001 posts

Re: GPT-5

#751

Post by morpheos137 »

highfrequency wrote: post #331

It is frequently suggested that once one of the AI companies reaches an AGI threshold, they will take off ahead of the rest. It's interesting to note that at least so far, the trend has been the opposite: as time goes on and the models get better, the performance of the different company's gets clustered closer together. Right now GPT-5, Claude Opus, Grok 4, Gemini 2.5 Pro all seem quite good across the board (ie the…

There is zero reason or evidence to believe AGI is close. In fact it is a good litmus test for someone's human intelligence whether they believe it.

What do you think AGI is?

How do we go from sentence composing chat bots to General Intelligence?

Is it even logical to talk about such a thing as abstract general intelligence when every form of intelligence we see in the real world is applied to specific goals as evolved behavioral technology refined through evolution?

When LLMs start undergoing spontaneous evolution then maybe it is nearer. But now they can't. Also there is so much more to intelligence than language. In fact many animals are shockingly intelligent but they can't regurgitate web scrapings.

Re: GPT-5

#752

Post by Aeolun »

I'm just sitting here hoping that their lowered prices will force Anthropic to follow suit xD

Re: GPT-5

#753

Post by Zambyte »

germandiago wrote: post #732
highfrequency wrote: post #331

It is frequently suggested that once one of the AI companies reaches an AGI threshold, they will take off ahead of the rest. It's interesting to note that at least so far, the trend has been the opposite: as time goes on and the models get better, the performance of the different company's gets clustered closer together. Right now GPT-5, Claude Opus, Grok 4, Gemini 2.5 Pro all seem quite good across the board (ie the…

Is AGI even possible? I am skeptical of that. I think they can get really good at many tasks and when used by a human expert in a field you can save lots of time and supervise and change things here and there, like sculpting. But I doubt we will ever see a fully autonomous, reliable AGI system.

No, AGI is not possible. It is perpetually defined as just beyond current capabilities.

Re: GPT-5

#754

Post by shruggedatlas »

lamontcg wrote: post #540
highfrequency wrote: post #331

It is frequently suggested that once one of the AI companies reaches an AGI threshold, they will take off ahead of the rest. It's interesting to note that at least so far, the trend has been the opposite: as time goes on and the models get better, the performance of the different company's gets clustered closer together. Right now GPT-5, Claude Opus, Grok 4, Gemini 2.5 Pro all seem quite good across the board (ie the…

> they can all basically solve moderately challenging math and coding problems Yesterday, Claude Opus 4.1 failed in trying to figure out that `-(1-alpha)` or `-1+alpha` is the same as `alpha-1`. We are still a little bit away from AGI.

Is this a specific example from their demo? I just tried it and Opus 4.1 is able to solve it.

Re: GPT-5

#755

Post by mickael-kerjean »

Buttons840 wrote: post #681
highfrequency wrote: post #331

It is frequently suggested that once one of the AI companies reaches an AGI threshold, they will take off ahead of the rest. It's interesting to note that at least so far, the trend has been the opposite: as time goes on and the models get better, the performance of the different company's gets clustered closer together. Right now GPT-5, Claude Opus, Grok 4, Gemini 2.5 Pro all seem quite good across the board (ie the…

The idea of singularity--that AI will improve itself--is that it assumes intelligence is an important part of improving AI. The AIs improve by gradient descent, still the same as ever. It's all basic math and a little calculus, and then making tiny tweaks to improve the model over and over and over. There's not a lot of room for intelligence to improve upon this. Nobody sits down and thinks really hard, and the resul…

I studied machine learning in 2012, gradient descent wasn't new back then either but it was 5 years before the "attention is all you need" paper. Progress might look continuous overall but if you zoom enough it might be a bit more discrete with breakthrough that must happen to jump the discrete parts, the question to me now is "How many papers like attention is all you need before a singularity?" I don't have that answer but let's not forget, until they released chat gpt, openAI was considered a joke by many people in the field who asserted their approach was a dead end.

Re: GPT-5

#756

Post by sbinnee »

I didn't know that OpenAI added what they call organization verification process for API calls for some models. While I haven't noticed this change at work using OpenAI models, when I wanted to try GPT-5 on my personal laptop, I came across this obnoxious verification issue.

It seems that it's all because that users can get thinking traces from API calls, and OpenAI wants to prevent other companies from distilling their models.

Although I don't think OpenAI will be threatened by a single user from Korea, I don't want to go through this process for many reasons. But who knows that this kind of verification process may become norm and users will have no ways to use frontier models. "If you want to use the most advanced AI models, verify yourself so that we can track you down when something bad happens". Is it what they are saying?

Re: GPT-5

#757

Post by piskov »

sbinnee wrote: post #756

I didn't know that OpenAI added what they call organization verification process for API calls for some models. While I haven't noticed this change at work using OpenAI models, when I wanted to try GPT-5 on my personal laptop, I came across this obnoxious verification issue. It seems that it's all because that users can get thinking traces from API calls, and OpenAI wants to prevent other companies from distilling th…

It started with o-models in the API.

Re: GPT-5

#759

Post by taylorlapeyre »

Telemakhos wrote: post #734

I am thoroughly unimpressed by GPT-5. It still can't compose iambic trimeters in ancient Greek with a proper penthemimeral cæsura, and it insists on providing totally incorrect scansion of the flawed lines it does compose. I corrected its metrical sins twice, which sent it into "thinking" mode until it finally returned a "Reasoning failed" error. There is no intelligence here: it's still just giving plausible output.…

It once again completely fails on an extremely simple test: look at a screenshot of sheet music, and tell me what the notes are. Producing a MIDI file for it (unsurprisingly) was far beyond its capabilities.

https://chatgpt.com/share/68954c9e-2f70-8000-99b9-b4abd69d1a...

This is not anywhere remotely close to general intelligence.

Post reply on HN