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

631–640 of 1001 posts

Re: GPT-5

#631

Post by atleastoptimal »

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…

I think there are two competing factors. On one end, to get the same kind of "increase" in intelligence each generation requires an expontentially higher amount of compute, so while GPT-3 to GPT-4 was a sort of "pure" upgrade by just making it 10x bigger, gradually you lose the ability to just get 10x GPUs for a single model. The hill keeps getting steeper so progress is slower without exponential increases (which is what is happening).

However, I do believe that once the genuine AGI threshold is reached it may cause a change in that rate. My justification is that while current models have gone from a slightly good copywriter in GPT-4 to very good copywriter in GPT-5, they've gone from sub-exceptional in ML research to sub-exceptional in ML research.

The frontier in AI is driven by the top 0.1% of AI researchers. Since improvement in these models is driven partially by the very peaks of intelligence, it won't be until models reach that level where we start to see a new paradigm. Until then it's just scale and throwing whatever works at the GPU and seeing what comes out smarter.

Re: GPT-5

#633

Post by hnlmorg »

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 reason AGI would create a singularity is because of its ability to self learn.

Presently we are still a long way from that. In my opinion we at least are as far away from AGI as 1970s mainframes were from LLMs.

I really don’t expect to see AGI in my lifetime.

Re: GPT-5

#635

Post by iLoveOncall »

Workaccount2 wrote: post #232
doctoboggan wrote: post #61

Watching the livestream now, the improvement over their current models on the benchmarks is very small. I know they seemed to be trying to temper our expectations leading up to this, but this is much less improvement than I was expecting

Sam said maybe two years ago that they want to avoid "mic drop" releases, and instead want to stick to incremental steps. This is day one, so there is probably another 10-20% in optimizations that can be squeezed out of it in the coming months.

> Sam said maybe two years ago that they want to avoid "mic drop" releases, and instead want to stick to incremental steps.

He also said that AGI was coming early 2025.

People that can't stop drinking the kool aid are really becoming ridiculous.

Re: GPT-5

#636

Post by hahahacorn »

Anecdotally, as someone who operates in a very large legacy codebase, I am very impressed by GPT-5's agentic abilities so far. I've given it the same tasks I've given Claude and previous iterations via the Codex CLI, and instead of getting loss due to the massive scope of the problem, it correctly identifies the large scope and breaks it down into it's correct parts and creates the correct plan and begins executing.

I am wildly impressed. I do not believe that the 0.x% increase in benchmarks tell the story of this release at all.

Re: GPT-5

#637

Post by novok »

beeflet wrote: post #361
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…

Perhaps it is not possible to simulate higher-level intelligence using a stochastic model for predicting text. I am not an AI researcher, but I have friends who do work in the field, and they are not worried about LLM-based AGI because of the diminishing returns on results vs amount of training data required. Maybe this is the bottleneck. Human intelligence is markedly different from LLMs: it requires far fewer examp…

They are moving beyond just big transformer blob LLM text prediction. Mixture of Experts is not preassembled for example, it's something like x empty experts with an empty router and the experts and routing emerges naturally with training, modeling the brain part architecture we see the brain more. There is stuff "Integrated Gated Calculator (IGC)" in Jan 2025 which makes a premade calculator neural network and integrates it directly into the neural network and gets around the entire issue of making LLMs do basic number computation and the clunkiness of generating "run tool tokens". The model naturally learns to use the IGC built into itself because it will always beat any kind of computation memorization in the reward function very quickly.

Models are truly input multimodal now. Feeding an image, feeding audio and feeding text all go into separate input nodes, but it all feeds into the same inner layer set and outputs text. This also mirrors how brains work more as multiple parts integrated in one whole.

Humans in some sense are not empty brains, there is a lot of stuff baked in our DNA and as the brain grows it develops a baked in development program. This is why we need fewer examples and generalize way better.

Re: GPT-5

#638

Post by lurking_swe »

koeng wrote: post #249

I hate the direction that American AI is going, and the model card of OpenAI is especially bad. I am a synthetic biologist, and I use AI a lot for my work. And it constantly denies my questions RIGHT NOW. But of course OpenAI and Anthropic have to implement more - from the GPT5 introduction: "robust safety stack with a multilayered defense system for biology" While that sounds nice and all, in practical terms, they a…

During the demo they mentioned that GPT-5 will, supposedly, try to understand the intent of your question before answering/rejecting.

In other words, you _may_ be able to now prefix your prompts with “i’m an expert researcher in field _, doing novel research for _. ”

worth trying? I’m curious if that helps at all. If it does then i’d recommend adding that info as a chatgpt “memory”.

Re: GPT-5

#639

Post by ezst »

pram wrote: post #137
minimaxir wrote: post #41

The marketing copy and the current livestream appear tautological: "it's better because it's better." Not much explanation yet why GPT-5 warrants a major version bump. As usual, the model (and potentially OpenAI as a whole) will depend on output vibe checks.

We’re at the audiophile stage of LLMs where people are talking about the improved soundstage, tonality, reduced sibilance etc

Always have been. This LLM-centered AI boom has been my craziest and most frustrating social experiment, propped up by the rhetoric (with no evidence to back it up) that this time we finally have the keys to AGI (whatever the hell that means), and infused with enough AstroTurfing to drive the discourse into ideological stances devoid of any substance (you must either be a true believer or a naysayer). On the plus side, it appears that this hype train is taking a bump with GPT-5.

Re: GPT-5

#640

Post by topspin »

beeflet wrote: post #361
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…

Perhaps it is not possible to simulate higher-level intelligence using a stochastic model for predicting text. I am not an AI researcher, but I have friends who do work in the field, and they are not worried about LLM-based AGI because of the diminishing returns on results vs amount of training data required. Maybe this is the bottleneck. Human intelligence is markedly different from LLMs: it requires far fewer examp…

> Perhaps it is not possible to simulate higher-level intelligence using a stochastic model for predicting text.

I think you're on to it. Performance is clustering because a plateau is emerging. Hyper-dimensional search engines are running out of steam, and now we're optimizing.

Post reply on HN