[dead]
GPT-5
861–870 of 1001 posts
Re: GPT-5
#862Some 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,…
I just don’t know that you’d name that 5. The jump from 3 to 4 was huge. There was an expectation for similar outputs here. Making it cheaper is a good goal - certainly - but they needed a huge marketing win too.
This tells me we're hitting a ceiling.
Re: GPT-5
#863It 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…
> It is frequently suggested that once one of the AI companies reaches an AGI threshold, they will take off ahead of the rest. This seems to be a result of using overly simplistic models of progress. A company makes a breakthrough, the next breakthrough requires exploring many more paths. It is much easier to catch up than find a breakthrough. Even if you get lucky and find the next breakthrough before everyone catch…
You are forgetting that we are talking about AI. That AI will be used to speed up progress on making next, better AI that will be used to speed up progress on making next, better AI that ...
Re: GPT-5
#864It 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…
It's also worth considering that past some threshold, it may be very difficult for us as users to discern which model is better. I don't think thats what's going on here, but we should be ready for it. For example, if you are an ELO 1000 chess player would you yourself be able to tell if Magnus Carlson or another grandmaster were better by playing them individually? To the extent that our AGI/SI metrics are based on…
> For example, if you are an ELO 1000 chess player would you yourself be able to tell if Magnus Carlson or another grandmaster were better by playing them individually?
Yes, because I'd get them to play each other?
Re: GPT-5
#865It 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…
Here's a pessimistic view: A hard take-off at this point might be entirely possible, but it would be like a small country with nuclear weapons launching an attack on a much more developed country without them. E.g. North Korea attacking South Korea. In such a situation an aggressor would wait to reveal anything until they had the power to obliterate everything ten times over.
If I were working in a job right now where I could see and guide and retrain these models daily, and realized I had a weapon of mass destruction on my hands that could War Games the Pentagon, I'd probably walk my discoveries back too. Knowing that an unbounded number of parallel discoveries were taking place.
It won't take AGI to take down our fragile democratic civilization premised on an informed electorate making decisions in their own interests. A flood of regurgitated LLM garbage is sufficient for that. But a scorched earth attack by AGI? Whoever has that horse in their stable will absolutely keep it locked up until the moment it's released.
Re: GPT-5
#866I've tried it in cursor and I didn't like it. The claude-4-sonnet gives me far better results.
Also it's a lot slower than Claude and Google models.
In general GPT models doesn't work well for me for both coding and general questions.
Re: GPT-5
#867First OpenAI video I've ever seen, the people in it all seem incompetent for some reason, like a grotesque version of apple employees from temu or something.
Re: GPT-5
#868It 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…
Here's a pessimistic view: A hard take-off at this point might be entirely possible, but it would be like a small country with nuclear weapons launching an attack on a much more developed country without them. E.g. North Korea attacking South Korea. In such a situation an aggressor would wait to reveal anything until they had the power to obliterate everything ten times over. If I were working in a job right now wher…
Pessimistic is just another way to spell 'realistic' in this case. None of these actors are doing it for the 'good of the world' despite their aggressive claims to the contrary.
Re: GPT-5
#869It 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…
In my experience and use case Grok is pretty much unusable when working with medium size codebases and systems design. ChatGPT has issues too but at least I have figured out a way around most of them, like asking for a progress and todo summary and uploading a zip file of my codebase to a new chat window say every 100 interactions, because speed degrades and hallucinations increase. Super Grok seems extremely bad at keeping context during very short interactions within a project even when providing it with a strong foundation via instructions. For example if the code name for a system or feature is called Jupiter, Grok will many times start talking about Jupiter the planet.
Re: GPT-5
#870Not impressed. gpt-5-nano gives noticeably worse results then o4-mini does. gpt-5 and gpt-5-mini are both behind the verification wall, and can stay there if they like.