[dead]
GPT-5
531–540 of 1001 posts
Re: GPT-5
#532It 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 feel like the benchmark suites need to include algorithmic efficiency. I.e can this thing solve your complex math or coding problem in 5000 gpus instead of 10000? 500? Maybe just 1 Mac mini?
Why? Cost is the only thing anyone will care about.
Re: GPT-5
#533So, would a layman notice the difference between GPT4 and GPT5 ?
Like a Turing test but between the models.
Re: GPT-5
#534I know HN isn’t the place to go for positive, uplifting commentary or optimism about technology - but I am truly excited for this release and grateful to all the team members who made it possible. What a great time to be alive.
I'm personally skeptical that the trajectory of this tech is going to match up to expectations but I agree HN has being feeling very unbalanced lately over it's reactions to these models.
Re: GPT-5
#535My 2 cents
There would be no GPT without Google, no Google without the WWW, no WWW without TCP/IP. This is why I believe calling it "AI" is a mistake or just for marketing, we should call all of them GPTs or search engines 2.0. This is the natural next step after you have indexed most of the web and collected most of the data.
Also there would be no coding agents without Free Software and Open-Source.
Re: GPT-5
#536It 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…
That being said, AGI is not a necessary requirement for AI to be totally world-changing
I think I want various forms of AI that are more focused on specific domains. I want AI tools, not companions or peers or (gulp) masters.
(Then again, people thought they wanted faster horses before they rolled out the Model T)
Re: GPT-5
#537Top 3 links in HN frontpage are all about GPT-5. I don't remember when was the last time people were so excited about something.
Re: GPT-5
#538so claude is doing so much thing before gpt 5 it's like a samsung vs iphone :D
Re: GPT-5
#539That SWE-bench chart with the mismatched bars (52.8% somehow appearing larger than 69.1%) was emblematic of the entire presentation - rushed and underwhelming. It's the kind of error that would get flagged in any internal review, yet here it is in a billion-dollar product launch. Combined with the Bernoulli effect demo confidently explaining how airplane wings work incorrectly (the equal transit time fallacy that NAS…
You're sort of glossing over the part where this can now be leveraged as a cost-efficient agentic model that performs better than o3. Nobody used o3 for sw agent tasks due to costs and speed, and this now substantially seems to both improve on o3 AND be significantly cheaper than Claude.
Re: GPT-5
#540It 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.