OpenAI simply can’t be trusted on any benchmarks: https://news.ycombinator.com/item?id=42761648
OpenAI claims gold-medal performance at IMO 2025
211–220 of 737 posts
Re: OpenAI claims gold-medal performance at IMO 2025
#212The cynicism/denial on HN about AI is exhausting. Half the comments are some weird form of explaining away the ever increasing performance of these models I've been reading this website for probably 15 years, its never been this bad. many threads are completely unreadable, all the actual educated takes are on X, its almost like there was a talent drain
Cynicism and denial are two very different things, and have very different causes and justifications. I personally don't deny that LLMs are very powerful and are capable and capable of eliminating many jobs. At the same time I'm very cynical about the rollout and push for AI. I don't see in any way as a push for a "better" society or towards some notion of progress, but rather an enthusiastic effort to disempower employees, centralize power, expand surveillance, increase profits, etc.
Re: OpenAI claims gold-medal performance at IMO 2025
#213The cynicism/denial on HN about AI is exhausting. Half the comments are some weird form of explaining away the ever increasing performance of these models I've been reading this website for probably 15 years, its never been this bad. many threads are completely unreadable, all the actual educated takes are on X, its almost like there was a talent drain
Meh. Some over hype, some under hype. People like you whine and then don't want to listen to any technical concerns.
Some of us are implementing things in relation to AI so we know it's not about "increasing performance of models" but actual about the right solution for the right problem.
If you think Twitter has "educated takes" then maybe go there and stop being pretentious schmuck over here.
Talent drain, lol. I'd much rather have skeptics and good tips than usernames, follows and social media engagement.
Re: OpenAI claims gold-medal performance at IMO 2025
#214Earlier quoted context omitted.
Probably because both sides have strong vested interests and it’s next to impossible to find a dispassionate point of view. The Pro AI crowd, VC, tech CEOs etc have strong incentive to claim humans are obsolete. Many tech employees see threats to their jobs and want to poopoo any way AI could be useful or competitive.
That's a huge hyperbole. I can assure you many people find the entire thing genuinely fascinating, without having any vested interest and without buying the hype.
Sure but it’s still a gold rush with a lot of exaggeration pushed by tech executives to acquire investors. There’s a lot of greed and fear to go around. I think LLMs are fascinating and cool myself having grown up with Eliza and crappy expert systems, but am more interested in deep learning outcomes like Alphafold than general purpose LLMs. You don’t hear enough about non-LLM AI because of all the money riding on LLM based tech. It’s hard not to see the bad behavior that has arisen due to all the money being thrown about. So that is to say it makes sense there is some skepticism as you can’t take what these companies say at face value. It’d be nice to have a toned down discussion about what LLMs can and can’t do but there is a lot of obfuscation and hype. Also there is the conversation about what they should or shouldn’t be doing which is completely fair to talk about.
Re: OpenAI claims gold-medal performance at IMO 2025
#215The cynicism/denial on HN about AI is exhausting. Half the comments are some weird form of explaining away the ever increasing performance of these models I've been reading this website for probably 15 years, its never been this bad. many threads are completely unreadable, all the actual educated takes are on X, its almost like there was a talent drain
> I've been reading this website for probably 15 years, its never been this bad... all the actual educated takes are on X Almost every technical comment on HN is wrong (see for example essentially all the discussion of Rust async, in which people keep making up silly claims that Rust maintainers then attempt to patiently explain are wrong). The idea that the "educated" takes are on X though... that's crazy talk.
With regard to AI & LLMs Twitter/x is actually the only place with all of the industry people discussing.
There are a bunch of great accounts to follow that are only really posting content to x.
Karpathy, nearcyan, kalomaze, all of the OpenAI researchers including the link this discussion is on, many anthropic researchers. It's such a meme that you see people discuss reading Twitter thread + paper because the thread gives useful additional context.
Hn still has great comment sections on maker style posts, on network stuff, but I no longer enjoy the discussions wrt AI here. It's too hyperbolic.
Re: OpenAI claims gold-medal performance at IMO 2025
#216Just a stochastic parrot bro
Re: OpenAI claims gold-medal performance at IMO 2025
#217The cynicism/denial on HN about AI is exhausting. Half the comments are some weird form of explaining away the ever increasing performance of these models I've been reading this website for probably 15 years, its never been this bad. many threads are completely unreadable, all the actual educated takes are on X, its almost like there was a talent drain
> I've been reading this website for probably 15 years, its never been this bad. People here were pretty skeptical about AlexNet, when it won the ImageNet challenge 13 years ago. https://news.ycombinator.com/item?id=4611830
That thread was skeptical, but it's still far more substantive than what you find here today.
Re: OpenAI claims gold-medal performance at IMO 2025
#218Earlier quoted context omitted.
They funded the entire benchmark and didn’t disclose their involvement. They then proceeded to make use of the benchmark while pretending like they weren’t affiliated with EpochAI. That’s a huge omission and more than enough reason to distrust their claims.
IMO their involvement is only an issue if they gained an advantage on the benchmark by it. If they didn't train on the test set then their gained advantage is minimal and I don't see a big problem with it nor do I see an obligation to disclose. Especially since there is a hold-out set that OpenAI doesn't have access to, which can detect any malfeasance.
There’s no way to figure out whether they gained an advantage. We have to trust their claims, which again, is an issue for me after finding out they already lied.
Re: OpenAI claims gold-medal performance at IMO 2025
#219Re: OpenAI claims gold-medal performance at IMO 2025
#220The cynicism/denial on HN about AI is exhausting. Half the comments are some weird form of explaining away the ever increasing performance of these models I've been reading this website for probably 15 years, its never been this bad. many threads are completely unreadable, all the actual educated takes are on X, its almost like there was a talent drain
It's caught in a kind of feedback loop. There are only so many times you can see "stochastic parrot" or "fancy autocomplete" or "can't draw hands" or "just a bunch of matmuls, it can't replicate the human soul" lines before you decide to just not engage. This leads to more of the content being exactly that, driving more people away. At this point, there are much better places to find technical discussion of AI, pros…
yeah alot of the time now, i will draft a comment, and then not even publish it. like whats the point