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

931–940 of 1001 posts

Re: GPT-5

#931

Post by BoorishBears »

neofytos wrote: post #927

For web app generation, gpt-5 seems to route my prompt to the reasoning model. Interestingly, gpt-4.1 often produced more creative or aesthetically interesting designs. In some cases, a bit of controlled hallucination (via temperature) is actually desirable, especially when generating UI ideas. Anyone else seeing this tradeoff?

All of these models generally produce hideous sites that because for some unknown reason we all decided shadcn and tailwind-with-no-design-system should define AI generated UI.

I use image generation for UI layout and have Claude implement it with an actual UI library: usually MUI with theming, but honestly anything with a sane grid system is better than loose tailwind.

Re: GPT-5

#933

Post by jeffnappi »

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…

We have no idea what AGI might look like, for example entirely possible that if/when that threshold is reached it will be power/compute constrained in such a way that it's impact is softened. My expectation is that open models will eventually meet or exceed the capability of proprietary models and to a degree that has already happened.

It's the systems around the models where the proprietary value lies.

Re: GPT-5

#934

Post by accrual »

Already a lot of comments here (2188 at the time of this comment) but wanted to share my 2c:

* It feels a bit more competent, as if it had more nuance or detail to say about each point.

* It got a few obscure details about OpenBSD correct right away - both Sonnet 4 and 4o sometimes conflate Linux and OpenBSD commands.

* It was fun asking GPT-5 to not only answer the query, but also to provide a brief analysis of the query itself for insights into myself!

Not a detailed review, but just a couple things I noticed with some limited usage.

Re: GPT-5

#935

Post by stoneyhrm1 »

somenameforme wrote: post #923
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…

If AGI is ever achieved, it would open the door to recursive self improvement that would presumably rapidly exceed human capability across any and all fields, including AI development. So the AI would be improving itself while simultaneously also making revolutionary breakthroughs in essentially all fields. And, for at least a while, it would also presumably be doing so at an exponentially increasing rate. But I thin…

> revolutionary breakthroughs in essentially all field

This doesn't really make sense outside computers. Since AI would be training itself, it needs to have the right answers, but as of now it doesn't really interact with the physical world. The most it could do is write code, and check things that have no room for interpretation, like speed, latency, percentage of errors, exceptions, etc.

But, what other fields would it do this in? How can it makes strives in biology, it can't dissect animals, it can't figure more out about plants that humans feed into the training data. Regarding math, math is human-defined. Humans said "addition does this", "this symbol means that", etc.

I just don't understand how AI could ever surpass anything human known before we live by the rules defined by us.

Re: GPT-5

#936

Post by whywhywhywhy »

sundarurfriend wrote: post #317

Some people have hypothesized that GPT-5 is actually about cost reduction and internal optimization for OpenAI, since there doesn't seem to be much of a leap forward, but another element that they seem to have focused on that'll probably make a huge difference to "normal" (non-tech) users is making precise and specifically worded prompts less necessary. They've mentioned improvements in that aspects a few times now,…

everything since the GPT-4 "Dev Day" downgrade from them has felt like cost reduction and internal optimization tbqh

Re: GPT-5

#937

Post by solumunus »

somenameforme wrote: post #923
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…

If AGI is ever achieved, it would open the door to recursive self improvement that would presumably rapidly exceed human capability across any and all fields, including AI development. So the AI would be improving itself while simultaneously also making revolutionary breakthroughs in essentially all fields. And, for at least a while, it would also presumably be doing so at an exponentially increasing rate. But I thin…

For some reason people pre suppose super intelligence into AGI. What if AGI had diminishing returns around human level intelligence? They still have to deal with all the same knowledge gaps we have.

Re: GPT-5

#938

Post by adhoc_slime »

surround wrote: post #297

GPT-5 knowledge cutoff: Sep 30, 2024 (10 months before release). Compare that to Gemini 2.5 Pro knowledge cutoff: Jan 2025 (3 months before release) Claude Opus 4.1: knowledge cutoff: Mar 2025 (4 months before release) https://platform.openai.com/docs/models/compare https://deepmind.google/models/gemini/pro/ https://docs.anthropic.com/en/docs/about-claude/models/overv...

funny result of this is that GPT5 doesn't understand the modern meaning of Vibe Coding (maximising llm code generation), it thinks it "a state where coding feels effortless, playful, and visually satisfying" and offers more content around adjusting IDE settings, and templating.

Re: GPT-5

#939

Post by mycall »

somenameforme wrote: post #923
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…

If AGI is ever achieved, it would open the door to recursive self improvement that would presumably rapidly exceed human capability across any and all fields, including AI development. So the AI would be improving itself while simultaneously also making revolutionary breakthroughs in essentially all fields. And, for at least a while, it would also presumably be doing so at an exponentially increasing rate. But I thin…

You are missing the point where synthetic data, deterministic tooling (written by AI) and new discoveries by each model generation feeds into the next model. This iteration is the key to going beyond human intelligence.

Re: GPT-5

#940

Post by Cthulhu_ »

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…

> Right now GPT-5, Claude Opus, Grok 4, Gemini 2.5 Pro all seem quite good across the board (ie they can all basically solve moderately challenging math and coding problems).

I wonder if that's because they have a lot of overlap in learning sets, algorithms used, but more importantly, whether they use the same benchmarks and optimize for them.

As the saying goes, once a metric (or benchmark score in this case) becomes a target, it ceases to be a valuable metric.

Post reply on HN