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

491–500 of 1001 posts

Re: GPT-5

#491

Post by netcan »

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…

Its certainly an interesting race to watch.

Part of the fun is that predictions get tested on short enough timescales to "experience" in a satisfying way.

Idk where that puts me, in my guess at "hard takeoff." I was reserved/skeptical about hard takeoff all along.

Even if LLMs had improved at a faster rate... I still think bottlenecks are inevitable.

That said... I do expect progress to happen in spurts anyway. It makes sense that companies of similar competence and resources get to a similar place.

The winner take all thing is a little forced. "Race to singularity" is the fun, rhetorical version of the investment case. The implied boring case is facebook, adwords, aws, apple, msft... IE the modern tech sector tends to create singular big winners... and therefore our pre-revenue market cap should be $1trn.

Re: GPT-5

#493

Post by gaptoothclan »

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…

I remember reading that llm’s have consumed the internet text data, I seem to remember there is an open data set for that too. Potential other sources of data would be images (probably already consumed) videos, YouTube must have such a large set of data to consume, perhaps Facebook or Instagram private content

But even with these it does not feel like AGI, that seems like the fusion reactor 20 years away argument, but instead this is coming in 2 years, but they have not even got the core technology of how to build AGI

Re: GPT-5

#494

Post by klik99 »

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’ve been saying for a while if AGI is possible it’s going to take another innovation and the transformer / LLM paradigm will plateau, and innovations are hard to time. I used to get downvoted for saying that years ago and now more people are realizing it. LLMs are awesome but there is a limit, most of the interesting things in the next years will be bolting more functionality and agent stuff, introspection like Anthropic is working on and smaller, less compute hungry specialized models. There’s still a lot to explore in this paradigm, but we’re getting diminishing returns on newer models, especially when you factor in cost

Re: GPT-5

#495

Post by jama211 »

hodgehog11 wrote: post #463
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

The hallucination benchmarks did show major improvement. We know existing benchmarks are nearly useless at this point. It's reliability that matters more.

I’m more worried about how they still confidently reason through things incorrectly all the time, which isn’t quite the same as hallucination, but it’s in a similar vein.

Re: GPT-5

#496

Post by chasd00 »

koonsolo wrote: post #451
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…

This confirms my suspicion that we are not at the exponential part of the curve, but the flattening one. It's easier to stay close to your competitors when everyone is at the flat curve of the innovation. The improvements they make are marginal. How long until the next AI breakthrough? Who can tell? Because last time it took decenia.

I think the breakthroughs now will be the application of LLMs to the rest of the world. Discovering use cases where LLMs really shine and applying them while learning and sharing the use cases where they do not.

Re: GPT-5

#497

Post by ricardobayes »

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…

AGI in 5/10 years is similar to "we won't have steering wheels in cars" or "we'll be asleep driving" in 5/10 years. Remember that? What happened to that? It looked so promising.

Re: GPT-5

#498

Post by jama211 »

Is it just me or has there not been a significant improvement in these models in the last 6 months - from the perspective of the average user. I mean, the last few years has seen INSANE improvement, but it really feels like it’s been slowing and plateauing for a while now…

Re: GPT-5

#499

Post by nerdix »

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…

Not only do I think there will not be a winner take all, I think it's very likely that the entire thing will be commoditized.

I think it's likely that we will eventually we hit a point of diminishing returns where the performance is good enough and marginal performance improvements aren't worth the high cost.

And over time, many models will reach "good enough" levels of performance including models that are open weight. And given even more time, these open weight models will be runnable on consumer level hardware. Eventually, they'll be runnable on super cheap consumer hardware (something more akin to a NPU than a $2000 RTX 5090). So your laptop in 2035 with specialize AI cores and 1TB of LPDDR10 ram is running GPT-7 level models without breaking a sweat. Maybe GPT-10 can solve some obscure math problem that your model can't but does it even matter? Would you pay for GPT-10 when running a GPT-7 level model does everything you need and is practically free?

The cloud providers will make money because there will still be a need for companies to host the models in a secure and reliable way. But a company whose main business strategy is developing the model? I'm not sure they will last without finding another way to add value.

Re: GPT-5

#500

Post by RealityVoid »

ricardobayes wrote: post #497
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…

AGI in 5/10 years is similar to "we won't have steering wheels in cars" or "we'll be asleep driving" in 5/10 years. Remember that? What happened to that? It looked so promising.

I mean, in certain US cities you can take a waymo right now. It seems that adage where we overestimate change in the short term and underestimate change in the long term fits right in here.
Post reply on HN