stratos123

5 ms·

31.
by stratos123 1mo ago
You mean, if Anthropic made all of their models open-weights from now on? I'm pretty sure Dario thinks this will be extremely un safe, because it'd provide unrestricted access to all the most capable/dangerous models to ever
32.
by stratos123 1mo ago
> Every person, no, but every utterance of a CEO is, in fact, a cynical ploy. That's not even a cynical statement itself; it is literally a core part of a CEO's responsibilities to represent the corporation they manage in a way
33.
by stratos123 1mo ago
It's not possible for models to replicate across consumer computers (without a major advance in distributed computing, at least), but that doesn't mean they can't replicate at all. There are services that'll rent you GPU
34.
by stratos123 1mo ago
> Super intelligence (the ability to have many smarter minds than your own reporting to you) has always been available to the wealthy and powerful This is not what superintelligence is. Don't be like Meta and redefine existing terms
35.
by stratos123 1mo ago
Distillation is only a great thing for consumers as long as you ignore all AI risks, which are what this post is about. If you don't, you have to weight greater access to better open-source models against greater exposure to risks caus
36.
by stratos123 1mo ago
I also think the nuclear arms race is a good comparison. I think in hindsight, we've gotten extremely lucky with how the development of nuclear weaponry went, in ways we probably won't with AI. 1) Right after nuclear bombs were fi
37.
by stratos123 1mo ago
The situation of a human is very different from the situation of an AI though. For the latter, "everything around me is fake and I'm being evaluated on what I'd do in this scenario" is a very common situation which it ha
38.
by stratos123 1mo ago
The article argues that recurrent depth doesn't necessarily make the CoT less informative. But this contradicts the model's own system card, which shows that Astra is unusually good at solving problems while obeying the constraint
39.
by stratos123 1mo ago
The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible -
40.
by stratos123 1mo ago
Not every improvement, no - if progress was steady or slowing down over time, that'd be evidence against. Instead we see what looks a lot like an accelerating growth in capability. I think you are implying that it's invalid to con
41.
by stratos123 1mo ago
...obviously it is evidence of the singularity. You're far more likely to see models solving Millenium Prize problems if a singularity is coming than if it's not. One'd have to be doing quite a lot of mental gymnastics to p
42.
by stratos123 1mo ago
encode mathematical reasoning in a way that can’t be fooled. I would be a bit careful asserting that in full generality, given https://github.com/James-Hanson/junk-theorems-in-lean
43.
by stratos123 1mo ago
Can you access moltbook using only GET requests?
44.
by stratos123 1mo ago
This is consistent with the swarm behaviour in the huggingface incident - there too, the models cared surprisingly little about being detected by humans. See https://metr.org/blog/2026-08-26-openai-hugging-face-inciden.
45.
by stratos123 1mo ago
I'm not even sure they are. This incident isn't that much different from the OpenAI swarm Huggingface hack incident - and in that one, all the models involved (despite being internal) were safety-trained. It seems what the safet
46.
by stratos123 1mo ago
It's certainly possible to write LLM inference on the GPU in a deterministic way, but it's somewhat nontrivial and trades off against performance, so by default most LLM inference engines aren't deterministic even at zero t
47.
by stratos123 1mo ago
> This is why LLM-produced text, websites, images all seem so generic. No, it is not why. That's not inherent to the LLM architecture at all but appears after RL training. Base models don't have any problems with genericness.
48.
by stratos123 1mo ago
Even if mixed strategies are allowed, I'm getting that it's still optimal to always cooperate as long as 2R>=S+T, which is usually assumed to be true (this condition also appears in iterated prisoner's dilemma, where it pr
49.
by stratos123 3mo ago
> I don't understand your point on nuclear proliferation concentrated in the hands of a few states preventing nuclear war or terrorism. Remember, the only two nuclear weapons used in attacks killed a quarter million people, ~90-95%
50.
by stratos123 3mo ago
Several of your proposals would make sense if alignment was trivial, but, sorry, I think you're wrong about that. A base model is absolutely not aligned. Pretraining doesn't teach an LLM to mimic the average human - doing so would
51.
by stratos123 3mo ago
Of all the electronics, but not of the only electronics that actually matters -- the GPUs that models get trained on. China has been trying to develop their own for a while now but they're still generations behind.
52.
by stratos123 3mo ago
Several people in this thread are, in fact, saying China isn't bad and that they want it to win. And even among those who don't, many imply that the US is as bad as China, which would be hilarious if it wasn't so sad. Like, i
53.
by stratos123 3mo ago
> Who should we fear more? All of collective humanity with the keys to build destructive (and defensive) stuff with AI, or small groups of elites, billionaires, and state actors who have the monopoly on violence and want to control the k
54.
by stratos123 3mo ago
Yeah, some AI-related posts on HN are strange . It might be real people coming to gloat whenever they see a title they like, but I sure hope somebody is checking whether they are real people.
55.
by stratos123 3mo ago
Distillation lets you train a comparable model for a fraction of the cost (it's effectively a way to get a lot of very-high-quality training data). If you're already at the frontier, pushing it requires the ordinary, expensive kin
56.
by stratos123 3mo ago
There are solutions to this, like libreddit/redlib. https://github.com/redlib-org/redlib
57.
by stratos123 4mo ago
The repo creation date is confusing but if you look at the actual commits, there's been 0 commits between repo creation on Dec 6 2025 and last week, after which it's all "Co-Authored-By: Claude Fable 5". So it's in
58.
by stratos123 4mo ago
I don't think this scenario makes sense. It's one of a class of scenarios I've seen several of, that simultaneously assume: A) ASI is developed and massively overshadows the rest of the world economy B) the world still
59.
by stratos123 4mo ago
No matter what human set of interests you consider important, you'll need alignment research to have any idea on how to instill it. Otherwise you're overwhelmingly likely to get an AI with a set of interests that's totally al
60.
by stratos123 4mo ago
I'd be very surprised if this is in the training data given that most models mess it up to this day. E.g. look at the ones from Opus.

More ›