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

591–600 of 1001 posts

Re: GPT-5

#591

Post by ACCount36 »

Therenas wrote: post #346
fidotron wrote: post #277

Going by the system card at: https://openai.com/index/gpt-5-system-card/ > GPT‑5 is a unified system . . . OK > . . . with a smart and fast model that answers most questions, a deeper reasoning model for harder problems, and a real-time router that quickly decides which model to use based on conversation type, complexity, tool needs, and explicit intent (for example, if you say “think hard about this” in the prompt).…

Too expensive maybe, or just not effective anymore as they used up any available training data. New data is generated slowly, and is massively poisoned with AI generated data, so it might be useless.

That's a lie people repeat because they want it to be true.

People evaluate dataset quality over time. There's no evidence that datasets from 2022 onwards perform any worse than ones from before 2022. There is some weak evidence of an opposite effect, causes unknown.

It's easy to make "model collapse" happen in lab conditions - but in real world circumstances, it fails to materialize.

Re: GPT-5

#592

Post by baxtr »

GolDDranks wrote: post #434
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 it's very fortunate, because I used to be an AI doomer. I still kinda am, but at least I'm now about 70% convinced that the current technological paradigm is not going to lead us to a short-term AI apocalypse. The fortunate thing is that we managed to invent an AI that is good at _copying us_ instead of being a truly maveric agent, which kinda limits it to the "average human" output. However, I still think th…

The AI dooming was never a thing for me. And I still don’t get it.

I don’t see anything that would even point into that direction.

Curious to understand where these thoughts are coming from

Re: GPT-5

#593

Post by WXLCKNO »

WD-42 wrote: post #219
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.

It has the last ~6 months worth of flavor of the month Javascript libraries in it's training set now, so it's "better at coding". How is this sustainable.

It doesn't even have that, knowledge cutoff is in 2024.

Re: GPT-5

#594

Post by KoolKat23 »

GolDDranks wrote: post #434
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 it's very fortunate, because I used to be an AI doomer. I still kinda am, but at least I'm now about 70% convinced that the current technological paradigm is not going to lead us to a short-term AI apocalypse. The fortunate thing is that we managed to invent an AI that is good at _copying us_ instead of being a truly maveric agent, which kinda limits it to the "average human" output. However, I still think th…

The only thing holding it back is lack of compute, and a lack of live world interface.

Re: GPT-5

#595

Post by kristianc »

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…

What I'm seeing is that as we get closer to supposed AGI, the models themsleves are getting less and less general. They're getting in fact more specific and clustered around high value use cases. It's kind of hard to see in this context what AGI is meant to mean.

Re: GPT-5

#596

Post by thimabi »

hrpnk wrote: post #336

They will retire lots of models: GPT-4o, GPT-4.1, GPT-4.5, GPT-4.1-mini, o4-mini, o4-mini-high, o3, o3-pro. https://help.openai.com/en/articles/6825453-chatgpt-release-... "If you open a conversation that used one of these models, ChatGPT will automatically switch it to the closest GPT-5 equivalent." - 4o, 4.1, 4.5, 4.1-mini, o4-mini, or o4-mini-high => GPT-5 - o3 => GPT-5-Thinking - o3-Pro => GPT-5-Pro

I personally hated this decision.

Of course, I know that having a line-up of tons of models is quite confusing. Yet I also believe users on the paid plan deserve more options.

As a paying user, I liked the ability to set which models to use each time, in particular switching between o4-mini and o4-mini-high.

Now they’ve deprecated this feature and I’m stuck with their base GPT-5 model or GPT-5 Thinking, which seems akin to o3 and thus has much smaller usage limits. Only God knows whether their routing will work as well as my previous system for selecting models.

Re: GPT-5

#597

Post by sigmoid10 »

hatthew wrote: post #503
fidotron wrote: post #277

Going by the system card at: https://openai.com/index/gpt-5-system-card/ > GPT‑5 is a unified system . . . OK > . . . with a smart and fast model that answers most questions, a deeper reasoning model for harder problems, and a real-time router that quickly decides which model to use based on conversation type, complexity, tool needs, and explicit intent (for example, if you say “think hard about this” in the prompt).…

I know this is just arguing semantics, but wouldn't you call it a unified system since it has a single interface that automatically interacts with different components? It's not a unified model , but it seems correct to call it a unified system .

It's not a unified architecture transformer, but it is a unified system for chatting.

Re: GPT-5

#599

Post by andrewinardeer »

Every release of every SOTA model is the same.

"It's like having a bunch of experts at your fingertips"

"Our most capable model ever"

"Complex reasoning and chain of thought"

Re: GPT-5

#600

Post by weego »

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'm still stuck at the bit where just throwing more and more data to make a very complex encyclopedia with an interesting search interface that tricks us into believing it's human-like gets us to AGI when we have no examples and thus no evidence or understanding of where the GI part comes from.

It's all just hyperbole to attract investment and shareholder value and the people peddling the idea of AGI as a tangible possibility are charlatans whose goals are not aligned with whatever people are convincing themselves are the goals.

Thr fact that so many engineers have fallen for it so completely is stunning to me and speaks volumes on the underlying health of our industry.

Post reply on HN