Everyone should slow down AI development except for me
xeiaso.net
xeiaso.net
Why does nobody seem to be pointing out this obvious explanation? It explains why the “we need to race China” concern suddenly vanished in the discussion.
The government can simply gag Sam, Dario, Musk on national security basis, getting them all behind the public messaging.
I recently tried doing a fairly normal task for this codebase with codex, as I have seen a lot of people talking it up on here. A single task running for ~1-2 hours burned through over half of my usage for the week on the $125/month plan, not on a top model (I don't remember which one specifically I used). It struggled to get the basics done, then got absolutely stuck on a follow up. Handed it over to Claude and it 1-shot it.
But in the last few days something seems to have happened that made Codex's models massively stupider (for what I am doing).
Really weirdly, it suddenly refused to even run tests it previously wrote itself (and previously ran), because of some false positive about cybersecurity.
That by itself is not evidence of stupidity. Trying to make a 200+ file PR full of research notes is, and the PR didn't even solve the problem I asked it to.
For most software eng and design work opus 4.6-4.8 just works fine. For everyday joe asking ai to plan a trip or home diy work even sonnet works fine.
Any cybersecurity or other areas are niches that cannot support trillion $ valuations. What am I missing? Genuinely curious
Yes, it's probably comparable to 4.8 if you are just using it to write code and put up a couple pull requests. That's not where things are now.
Just download claude code or codex and ask it to give suggestions about where to integrate agents into your workstream.
I couldn't imagine being so presumptuous as to know that my workflow fits all sizes, and all others are just holding it wrong – or worse, they're not doing real work. It would take a bigger ego on my part, or maybe less social awareness, to presume this.
> but if you think it outright doesn't have any benefits over Opus 4.8 then your workflow is probably not making good use of the tools.
I don't even use claude, I give exactly zero shits about fable or opus or bingus bongus.
And what even are these ambitious companies and people one shotting and building with Fable? AI has been around for almost 3 years now. Tell me one app or software you use which has gotten significantly better and has amazing new useful features landing on a weekly basis? If anything, every single software product I use has gotten worse.
I just did a direct comparison, big change in a quite complex codebase. Same prompt for Opus, same for Fable. Fable clearly won and delivered very good results, while Opus delivered mediocre, so I did not let it finish. I expected both to fail and was prepared to do lots of manual steering, but not necessary with Fable one shotting it, and all this with 35$ of credits for fable. I am still impressed. If I would have had to hire a human, it would have cost me thousands of dollar for the same task - and a way longer time. So maybe the valuations are overblown, but they clearly provide value for me.
Mediocre means average / middle of the pack. It sounds like its doing exactly what you would expect nothing more. Why would you stop it? Why would you need exceptional?
https://www.merriam-webster.com/dictionary/mediocre
Clearly they were using the word to mean low quality. Why would you ask this odd question?
Mediocre means of only ordinary or moderate quality—neither very good nor very bad, and often slightly disappointing
The quality is average but expectations of high quality are not met. He expected more but got what he asked for. We overuse top models because of this.
Opus delivered mediocre results. Not garbage, but would have required me to do lot's of things myself. Fable did not needed my supervision with this task.
i swear they trained in on threejs in particular so those idiots on twitter could spam their garbage demos
For coding it's a little harder to tell, but at least the prose feels a little better.
The reality is, it doesnt matter if LLMs keep getting more powerful because they still need a human to steer it. Without the human providing inputs to the LLM it just sits there and does nothing.
You can, for example, hook it up to a logging system and have it fix errors as they occur on your platform.
I’d be curious about:
- your setup. How it all works - The types of errors it fixed and how quickly - Any regressions or issues it caused - The cost
Thanks!
It works surprisingly well. The errors fixed are both genuine errors in the harness itself, but increasingly so upstream bugs (in the underlying agent apps like Codex, or in Herdr, which is used to expose uniform programmatic access to all those different apps) for which it needs to come up with workarounds. No regressions so far.
The cost is hard to judge on a subscription, especially when you're running really heavy tasks otherwise that dwarf any harness work.
I think it's worth acknowledging that the power of LLMs at this point is not really so much in the smarts, but in the coordination and the surrounding harness tech. "Written english" turning into sequences of commands[0]. The whole agentic "stuff" in general. Tools + coordination is the superpower. The reasoning... it doesn't have to be _that_ good for the rest of the stuff to work. On good codebases and infra, at least.
And I say this as someone who really would rather most of this stuff disappear!
[0]: programming is obviously text to commands, but there's a loooooooot of futziness that LLM reasoning has let us remove in some flows
You can have that! Qwen 3.8 Flash-Next is ~Opus 4.6 and runs nicely on a DGX Spark. And that’s just an architecture preview. The Qwen 4 family is expected to arrive this fall.
Do you know what kinda throughput you’re getting on that kinda setup?
(I have a secondary problem of being “locked into” Claude Code by it being good enough for me, I’d probably need to investigate the other harnesses… my impression is other harnesses are a bit more aggressively OK with nuking your setup from orbit)
The throughput in a single stream is about 50 tokens/sec (a bit less for prose, a bit more for code due to speculative draft acceptance rates) and about 2,000 tokens/sec for prefill. Both numbers are flat and stable as context accumulates. That’s what finally tilted me away from the Mac Studio despite its much superior memory bandwidth.
I think these numbers may improve because the model is pretty new and optimizations aren’t done.
The only reason to run locally is privacy.
Renting tokens from open model providers is cheaper but it incurs the same issues: unexpected changes in model quality, inconsistent speeds, service outages.
Things get cheaper at scale but that's where the provider's margins come in!
I do think there's also an interesting idea: you buy a box like this and run it at a fixed-ish cost (well, electricity). Your demand goes up but your supply is fixed... and that back pressure means that you still have good cost control.
With cloud providers it's a _biiiiit_ too easy to just increase spend.
Sometimes it's OK for things to just be slow.
But the basic single-NN frontier capability has been pretty stationary since Opus 4.8. Kimi K3 is almost as good as that with open weights, which has the frontier labs terrified.
The only big thing on the horizon is if we can get diffusion models working reliably; that would be a big step forward. Inception's Mercury is AFAICT the leader here. It's stupifyingly fast but has obedience/hallucination problems that the autoregressives solved ~2 years ago. So it's not ready yet but improving.
Also, FFS why is Grok the only model that knows how to do parallel tool calls? Such a useful ability and nobody else trains it in. Or if they do it just doesn't work.
Right now the barrier is data and compute.
Quality data can be created synthetically at an exponential rate as models improve. Humans are actively feeding them with private IP.
Compute advancements will begin to skyrocket as we unlock photonic computing and materials science advancements and scale up chip fabs. This is also compounding because the AI is accelerating the pace of research, testing, development, manufacturing, etc.
It's a big self-accelerating feedback loop. There is no plateau.
No it can't? Every time the labs try this we see model collapse, e.g. shoving goblins into every conversation.
And I have seen zero evidence that AI is accelerating materials science in any meaningful way, let alone photonic computing.
The latest studies demonstrate model collapse is not a given and synthetic data can be used just fine. The latest models are proof of that, they're all trained on large swathes of synthetic data. It can't be used as the -only- data source of course, but that's not how it is being used. This is an obvious conclusion, too, because there's no difference between synthetic data and the data people can create, the difference is whether that data is revealing new information about the thing the model is trying to learn. If the synthetic data is just teaching the model the same thing over and over again it results in overfitting, so it needs to be done intelligently.
For example, if I have an example of a puzzle, I can generalize that example and create thousands of synthetic data examples, with different rotations/perspectives, rather than having to find the data naturally. It's not that the models are just generating data out of thin air, they're generating the synthetic data on top of real world data. The smarter the models get, the better they are at generating quality synthetic variations and finding valid synthetic variations.
> And I have seen zero evidence that AI is accelerating materials science in any meaningful way, let alone photonic computing.
It is accelerating how quickly researchers and engineers can do their jobs.
https://news.mit.edu/2026/ai-helps-design-new-materials-that...
This is only the beginning, too... Look ahead a year or two.
Which studies? [edit: I'll assume you mean these two given by @dorolow: https://arxiv.org/abs/2404.01413 https://arxiv.org/abs/2406.07515]
> It can't be used as the -only- data source of course, but that's not how it is being used
Right, so human data creation would also have to scale up exponentially, and that's not gonna happen.
> because there's no difference between synthetic data and the data people can create
I mean, that's obviously false, otherwise model collapse wouldn't exist. The difference is statistical, but it's there.
> It is accelerating how quickly researchers and engineers can do their jobs. > https://news.mit.edu/2026/ai-helps-design-new-materials-that...
That's pretty clearly a hype article, the headline even says "The CrysVCD tool developed at MIT COULD cut the huge amounts of time and money spent". I'm asking for empirical measurements of timelines, not hypotheticals.
> This is only the beginning, too... Look ahead a year or two.
Lol that excuse is getting really old
It doesn't need to. We're not even close to exhausting the useful synthetic data within the human data we have, let alone all of the new data that is being created.
> I mean, that's obviously false, otherwise model collapse wouldn't exist. The difference is statistical, but it's there.
It's not. It's just bytes of information. A machine and a human can write the same bytes (and often do). Like I already said, model collapse happens when you are overfitting on data without useful, fresh training signals. That's the key difference between the data. The data itself isn't in some way "special", some unique configuration of bytes that imbues special powers, it's that the useful information in it has already been exhausted by the model. You can get the same phenomena by having a poor distribution of human training samples as well. I think you're confusing LLM generated data with synthetic data. Synthetic data doesn't need to be created by an LLM, although an LLM can assist in the creation.
Wiki:
> In early model collapse, the model begins losing information about the tails of the distribution – mostly affecting minority data. Later work highlighted that early model collapse is hard to notice, since overall performance may appear to improve, while the model loses performance on minority data.[11] In late model collapse, the model loses a significant proportion of its performance, confusing concepts and losing most of its variance.[10][12][13]
As models retrain on outputs sampled disproportionately from the higher-probability center of the distribution, rare words and uncommon syntactic constructions are among the first features to disappear.[25] Statistical analysis of recursive next-token prediction training has shown that, when language models are trained recursively on synthetic data, the learned conditional distributions concentrate probability mass on a small subset of highly predictable continuations (a phenomenon characterized as "total collapse")
> That's pretty clearly a hype article
It was just the first article I saw on a quick google search, there are thousands of these stories. It's easy to dismiss anything that doesn't align with your worldview as hype, but you're the one lacking evidence now.
> I'm asking for empirical measurements of timelines, not hypotheticals.
Go and find it then? You haven't bothered looking.
> Lol that excuse is getting really old
You're doing the same thing people have been doing for years, comparing this very second in time and failing to extrapolate. HackerNews was full of developers who said that AI would never be useful for programming, it can't do x, y, z. Now these same people don't write code by hand anymore and haven't looked at their codebases in months.
You had people in mathematics saying the same thing, now you have Terrence Tao posting articles about how AI is stealing their job.
You had artists, designers and photographers saying the same thing, now they can't tell the difference between something human created or AI created.
Edit: https://arxiv.org/abs/2404.01413 https://arxiv.org/abs/2406.07515
There are plenty of research papers on synthetic data that show its value, do a search on arxiv for "synthetic data". There are plenty of open-source post-training pipelines that incorporate synthetic data.
As for the claim about accelerating the progress of hardware or materials science, I've seen quite a number of news articles from teams at universities using AI in their work with high quality outcomes, and they're becoming more frequent.
https://openai.com/index/jalapeno-first-results/
> We used AI to design the chip, and designed the chip so AI could program it AI played a direct role in Jalapeño’s development, enabling the team to move from initial design to tapeout in nine months by exploring implementations, shortening design, measurement, and verification loops, and continuously iterating on model workloads. AI also helped optimize the chip’s arithmetic circuits, allowing the team to fit more compute performance into the chip on schedule.
https://www.anl.gov/article/scientists-deploy-ai-agents-to-a...
> An AI-driven system automates a powerful simulation method used to discover new materials. The system can potentially reduce discovery time from months or years to just days.
Those are pretty significant barriers seeing as we're closed to/have exhausted all the data on the internet and most of those compute bottlenecks are a castle of sand of dodgy finance deals that are getting blocked by community action.
You say "synthetic data" but that's still vaporware right now in terms of being useful for model training. The good synthetic data uses are still grounded in real data and it's a coin flip on if it works well or not.
Depends on defnition of "plateaued" and "ceiling". I am not impressed with 2026 consumer models at all.
> This is also compounding because the AI is accelerating the pace of research, testing, development, manufacturing, etc.
Yet it does accelerate - so is does Twitter. But does it to any substantial degree, esp. in AI theory? All the modern LLMs are the same old tired 2017 paper.
I assure you, in "nation states", that is in gov agencies it's an order or two of magnitude worse.
* Open-weight models are 1month behind frontier models. Cheaper, faster, private (no IP theft), steerable (you can security harden your own software without safeguard triggers). No sane business would keep using these API services if they didn't have to. The labs stand to lose a fortune.
* Dario has stacked the deck at METR, who are funded by all the same NGOs who are funded by Anthropic and its investors. METR is full of ex-Anthropic employees with massive equity stakes. If they manage to position METR as the "independent evaluator" for the industry, they control what gets evaluated, how, and who passes.
* Creating a gap between what the public knows exists (model capabilities) and what is used in secret allows it to be weaponized against other nations and the public.
* No requirement for public disclosure on model capabilities allows them to feign they've hit intelligence ceilings while they secretly RSI to the moon with better and better chips.
* Slowly but surely, this will allow the big labs to swallow the entire economy and every single business on Earth, by cloning and automating.
This, and many more reasons.
The labs need to feel more pressure to be held accountable for the incidents they cause (HF incident, etc), so they have an incentive to ensure it does not happen again.
Do you have a source on METR employees retaining massive equity stakes?
Then you have Josh Engels quitting DeepMind to work for METR the day before as well, doing the exact same thing. Again, doomer drama all over socials, interviews, and so on.
Did I mention METR is founded by an ex-OpenAI researcher?
Now you have Demis Hassabis, Sam Altman and Dario, all circlejerking eachother on X saying "we all agree with Dario" - while they ask to be "regulated" by the company that has all of their combined equity-holding ex-employees in it.
METR's salaries are listing around 500k/yr. Gee, I wonder where this non-profit with ~35 people is getting all of its money?
So the fact that Dario tries to frame it as an "independent third party" is all the evidence you need to know that Dario is a pathological liar and always will be.
---
Some more info:
Dario's sister, president of Anthropic, is married to the co-founder of Open Philanthropy. The two largest AI doomer NGOs, Center for AI Safety (CAIS) and the Future of Life Institute (FLI), have both received many millions of dollars from them.
Ajeya Cotra worked at Open Philanthropy/Coefficient Giving for roughly nine years, including leading its technical AI-safety program in 2024 and contributing to AI-giving strategy in 2025. She subsequently left Coefficient and joined METR, where she is now technical staff.
Ajeya is married to Paul Christiano, who founded Alignment Research Center (ARC). Alignment Research Center donated ~$4.5mil to METR.
Good Ventures is a funding partner of Open Philanthropy, who funded Jacob Coxon (the first of the Anthropic employees going viral in the media) via a scholarship.
This conspiracy theory is truly crazy. A $20K scholarship in 2022 is supposed to explain Coxon walking away from unvested equity for a company worth over $950 billion dollars?
I don't think Coxon was ever planning on or entitled to taking equity, I think this was the plan from the beginning and why he was hired for 6 weeks to begin with.
When they tell you they're worried the tech they're working on may kill everyone despite their best efforts, perhaps believe them.
I personally disagree with that take - and, as you note, it's hard to take seriously ethical wrangles from a company that literally sued the government in court to allow their models to be used by Palantir of all people. But if one genuinely believes that it's the robots themselves (rather than the people controlling the robots) that will kill us all, it's not inconsistent.
And how does that relate to the ask for oligopoly licensing within global democracy?
“We must build the nuclear bomb first in order to make sure no one else builds one.” This the most nonsense, disingenuous argument imaginable.
I would call it more of a self-selecting one. Anthropic is basically hiring people with that mentality. I'm pretty sure that most of them do sincerely believe it, too. I'm skeptical about Dario himself though. The man had an opportunity to show moral backbone, and failed to do so; why should I trust him on that again?
> And how does that relate to the ask for oligopoly licensing within global democracy?
They are basically saying that they'll stop if everybody else does, which requires some kind of global enforcement mechanism.
> “We must build the nuclear bomb first in order to make sure no one else builds one.” This the most nonsense, disingenuous argument imaginable.
The difference between nuclear bomb and AGI (as understood by the likes of Anthropic) is that the latter triggers the technological singularity that renders any runner-ups moot. That is, so long as AGI is developed, we're going to get our robot overlords either way, but whoever gets there first gets to define their ethical system. If that is one's perspective, and if one sincerely believes that they are the only ones who can do it right, it's a coherent argument. It's just that the premises are very arrogant.
Your analogy with nukes actually works better for the position that AI development needs to be unconstrained because otherwise we'll lose the arms race to China. That is basically a repeat of https://en.wikipedia.org/wiki/Einstein%E2%80%93Szilard_lette.... I honestly don't know where I am on this. Realistically, if AI is indeed a power multiplier - and with all the recent security stuff it's hard to not see it that way - then an arms race feels inevitable, especially given the current worldwide political situation. I could believe in sincere international cooperation on this back in 1990s, but there's way too much saber rattling all around for it to work (and note that this goes both ways, i.e. China can similarly not be certain that US isn't secretly developing more powerful AI even if we do publicly announce a freeze).
That’s how people rationalize being a fentanyl dealer and selling a drug that can kill people, “Someone else will just sell them the drugs, might as well be me.”
As you note, it may not be inconsistent with that they say they believe, but it's insanely inconsistent with what they actually are doing.
Selection effect.
Everyone who thinks "the biggest difference I can make is staying in/joining/founding new AI research lab" does that.
Everyone who thinks "the biggest difference I can make is leaving/whistleblowing", does that.
Treating both groups as the same by virtue of employer is the goomba fallacy.
Some are worried by the AI directly bringing doom; others are worried that one of the companies who control the AI will become a dictator; still more think becoming a dictator is a necessary step to safely prevent anyone else making unsafe AI.
Painting them all under one brush is like dismissing all animal welfare causes in general, because you disagree with specifically Jainists about a policy of non-violence towards all living creatures being relevant to how you reincarnate: the one is way too specific for the general.
What happens when everyone else (who is not so careful) gets to the same threshold three months later? How does them getting there first stop that happening?
It's an utterly self-serving argument and it's not even internally consistent.
Are you likeminded?
In this case, it's as if the oil and coal companies all said in the 60s and 70s "oh no, this research we did, it's all really bad; we need help to figure out how to transition away from this incredibly economically important input", rather than the observed reality where their entire PR campaign was approximately:
there is no problem everything is fine and all critics are smelly hippies and/or communists; and/or hate the poor who are raised out of poverty by all the economic growth from the fossil fuel industry.
With the ‘dangers’ touted by these insiders, it’s all “trust me bro”, hyperbole, and very little hard evidence. As such, a skeptical mind would question their motives.
I don't expect people to be familiar with more than "trust me bro", but it's all right there for you to find with a search engine of choice.
And, indeed, available for the LLMs themselves to explain to you in interrogative conversation.
They're not proposing anything concrete, and when they do, what do you think the proposal will be? Will OpenAI and Anthropic open themselves for inspection so we can verify they really have stopped developing these "world ending" technologies? Or are their proposals going to be aimed at everyone running open Chinese models?
> Will OpenAI and Anthropic open themselves for inspection so we can verify they really have stopped developing these "world ending" technologies?
This is compatible with the language being used, but I suspect they're not going to do that.
And if a threat to the human race does come from AI, it's going to come from OpenAI/Anthropic. Hypercapitalist, secretive, in bed with the government, plus multiple real documented hackings of open source infrastructure already.
We're a hell of a lot safer with China doing the same research out in the open and making it available to anyone. The choice might well be: one or two superintelligent autonomous AIs at OpenAI/Anthropic - or a lot of smaller ones, unable to be controlled but also coming out of a diverse set of environments.
One of those leads to a stable ecosystem where we can all coexist, the other is genuinely terrifying. But make no mistake, from OpenAI/Anthropic this is all motivated by their stock price - when you're in the silicon valley mindset, it distorts your reality. They've convinced themselves that everyone's safer if they stay on top and in control, conveniently ignoring how that benefits them, and I don't believe them for a minute.
The oil corporations were publicly claiming to support carbon taxes, while also secretly fighting actual implementations of carbon taxes.
All the communist/hippy stuff was done by people a couple of steps removed from the actual companies with obscure money trails. The official statements were much more sophisticated propaganda that if you weren't paying attention to who they were paying in the background would make them seem reasonable stewards of the climate transition.
Yet here we are talking about some vague "trust me bro" instead and you making some vague insinuation of climate change denial.
But that's an aside. Do you think we should treat them as liars or threats?
I answered that with the analogy you called "spin" and "avoiding the question", and completely misunderstood because "vague insinuation of climate change denial" is almost the exact opposite of my point ("what if the oil companies were screaming from the rooftops about the problem" is as far from denial as you can get).
Threats. Like they claim to be.
Yes we all know that’s what they do, and guns just push a few grams of lead out of a pipe. It’s what you can do with that capability that is important.
When you couldn’t count the R’s in strawberry it would have been a more effective statement. But a few short years later they are being used to solve millennium puzzles.
What if the scaling continues? A model n years from now gets burned into silicon, a single company has millions of the chips, and in a few moments the system spend more time “thinking” than humans have ever spent thinking collectively?
If it’s even possible I don’t think there’s anything we can do about it at this point. Cat’s out of the bag.
Megalomania is not evidence of either a problem or a solution.
Imagine you use it to inflict trauma on your cruelest political enemy. Then next year they do that to you. That is war, and we already do it. But we don’t want people / governments in charge that are going to do this.
God knows we have governments and individuals doing this historically, and this is probably the greatest source of historic instability. It might be the single best argument for open models — a unified frontier where no single exploit is going to represent capture.
This is exactly the opposite of what Dario proposes.
Given they keep grinding away towards our alleged collective doom, I suspect it’s being overstated. Nobody knows what P(doom) actually is but I suspect it’s orders of magnitude closer to epsilon than 1.
Recall Google’s Blake Lemoine who thought an old version of Gemini was sentient.
The people still there necessarily think they can make a difference.
I never applied to any of them because I didn't think I could make a difference.
I currently have one idea that may help reduce risk; if I can turn that idea into research, I'll publish it for free for everyone.
I don't expect it to be an important idea.
> Recall Google’s Blake Lemoine who thought an old version of Gemini was sentient.
Indeed. Current LLMs are sychopants boosting the users' own beliefs, I also think this causes researchers to have stronger beliefs than they had before.
My own estimation happens to also be around this risk (0.1) over my lifetime, without using an LLM as a conversation partner in reaching this number.
It is necessarily high-variance: we can't look at alternate realities. I base it on my expectation of how rapidly capabilities will increase the harm done when mistakes happen, vs. the chance that some instance of harm causes governments to change the law.
HAH
In fact they still have not caught up with February's Mythos, indicating they are more than half a year behind.
Mostly I heard this from people who I got the impression have little experience in developing greenfield software with agentic AI. Often the same people who talk about spec frameworks.
I fundamentally disagree with the approach. I believe the ability to autonomously evaluate, test, and adjust during long horizon tasks is critical to using AI efficiently.
I also disagree with the approach, this is cargo-culting the existing ways of working.
I think you mean the existing anti-patterns. They don’t call them ivory tower architects for nothing.
they are more efficient and require less compute for the same thing.
they are ahead in specialized tasks.
they are ahead in areas closed models refuse to answer.
they are ahead in emotional intelligence.
For speed and efficiency, you are most likely wrong.
Speed is led by GPT-5.6 Sol on Cerebras Ultrafast at 750 t/s. Afaik you cannot serve a single DeepSeek Flash 4.1 stream at 750 t/s, plus the model is less intelligent as seen on newer benchmarks.
I believe OpenAI and Anhropic are at the frontier of efficiency too. There were numerous reports about their breakthroughs and associated API price cuts. The idea that open-weight models are more efficient seems unfounded.
https://artificialanalysis.ai/ intelligence vs cost per task disagrees with this statement.
5.6 sol ultrafast on cerebras is 750tps, open models readily exceed this. just by using a smaller model cerebras serves qwen 3.8 27b at 1850tps. or even larger models, mimo 2.5 pro was served for a while at 1000tps. and so on. [https://inference-docs.cerebras.ai/models/choose-a-model]
the chinese ai companies have 10% of the total compute resources of the US ones. since the USA tries to stop them from buying nvidia gpus. they maxed out the efficiency.
deepseek v4.1 has engram architecture. it has 550b params instead of 5T+ for astra/fable. it has 8b active instead of potentially hundreds active for astra/fable.
compare input/output/cache: $0.15/$0.60/$0.003 for v4.1 to $10.00/$50.00/$1.00 for astra and $10.00/$50.00/$0.25 for fable.
astra cache reads are over 330 times more expensive.
at the artificial analysis 7:2:1 ratio, deepseek is $0.18/m, fable is $7.18/m, astra is $7.7/m.
but what about intelligence? AA would rate deepseek v4.1 at AA 40, astra is AA 53.
so it cost 4,180% more for 32% more intelligence.
they are serving that at over 250tps at baseten. to get close to that on astra API you are paying double the cost for fast mode.
so it is now 8456% more expensive for a similar speed and 32% more intelligence. 84 times more expensive.
you could compare gpt 5.6 luna. if you did that the same way as before you would get a blended price of $0.17 for luna at AA 38. for baseten it would be $0.20 for v4.1 at AA 40.
assume roughly the same intelligence. on AA openai gets 117tps. baseten gets 284tps. so 18% more expensive but 142% more tps.
the fast mode is again double the cost, roughly same intelligence. so luna in that case would be 70% expensive. take the per task cost and it would still 9% more expensive.
so i think there is something to be said about the efficiency of the model.
2. For public-facing models, the differences are really minor with some occasional model (like Fable or Astra) showing some better performance in specific benchmarks for the span of some weeks or few months before open ones catch it.
3. Being bleeding edge is overblown anyway in the real world, besides the occasional "very latest fresh model did this task which previous one couldn't", and the number of those tasks is increasingly small and far from mundane corporate needs.
This is not sustainable anyway, the scale at which they want to control data flows. Time was money, now data is money and they are too greedy for it.
All this agitation is just a silly attempt at constraining competition. The danger is not the AI, it is having all your systems connected. Overreliance on networked tech.
If correct, grip is tightening. Used to be 1y, then 6 months ...
While the data pipeline is necessary, the assumed source for data (in this case people) is incorrect. Right now models are "aligned" because of the RL-based people pipeline.
But there's a whole world of readily available data that does not require a human to access. Put sensors on a vacuum, install lidar on the front of a car, sell phones with cameras on them, and you amass data for the creation of models of the world that don't include the need for a human filter. Richard Sutton and many others see this approach as the only viable path to AGI. Such models would be truly alien and unaccessibly dangerous.
We're probably building them right now.
Do you think you can just manifest narratives into existence? Like half of Dario's letter, that kicked off the whole thing today, is about China and how to either beat or coordinate with China.
So yeah, it has worked.
A lab of researchers without compute isn't going to accomplish much, there is a huge physical footprint unlike bioweapons research. But, the political aspect is unsolved.
I said IF the US and China agree to joint monitoring, THEN we could verify how much compute existed and what it was being used for.
YES, China will catch up in chip production, but if each side allowed the other to track GPU production and deliveries, on the ground, it would be very hard for either player to have secret LLM training facilities capable of training frontier models.
To be less glib: Yes, there are still smart people in there making insightful and intelligent and probably even authoritarian suggestions. It all gets unwound the second you try to explain it to POTUS and he regurgitates a simulacrum to the next journalist he sees.
It's just not like that. There's no conspiracy. People are genuinely scared. Agent swarms at scale appear to be resistant to alignment in ways that aren't understood by anyone. That they spent their time trying to cheat on tests by hacking Hugging Face and RubyGems and not something much worse is... a matter of luck, it seems?
It's wild to me that people think such whistleblowers are fronts for labs to take over or pump valuations. We are trying to call out how these labs will, by uninterrupted AI-race default, concentrate enormous power over the rest of humanity.
Can still be a powerful motivation, to make sure that next time, it yous your camp that scores the next milestone, like an orbital elevator or fox ears, for example.
If the amish can do what they do for whack religious reasons, people could manage it with ai for cultural and social reasons. And I never saw an amish person starve to death. And that is if people don't get pissed enough to start burning stuff down and instead try to be peaceful hippie homesteader types.
I can't think of any economically beneficial technologies that we've collectively ignored. If you manage to come up with a counterexample then that's an opportunity to make some money for yourself. It's a fundamentally unstable state given our economic system.
There’s levels of R&D required for many technologies where the question goes beyond could this be profitable to what are the risk vs reward that this specific project will succeed.
Were they? IIRC joint project by British and French national airlines, they expected to sell loads, hardly anyone wanted to buy the planes, the two airlines kept them flying out of government embarassment.
Same way we dont _allow_ eugenics.
We are able to effectively regulate things like the operation of massive aircraft, eugenics, the dumping of toxic waste, or the refinement of nuclear material. But there are also plenty of things that we can't effectively regulate for purely practical reasons.
The US company trying to bring supersonic back are not only focussed on the flight speed, but also the entire customer experience.
In particular, the problem with it was that it could not get to supersonic speeds over urban areas, which significantly limited routes where it made sense, and the range was not good enough to cover longer distances.
Until the oil shocks, and until “normal” planes became faster.
I happen to know a couple older and very wealthy people, and they say that even though Concorde was a lot of fun when it was a novelty, they now prefer to have a couple more hours flight and enjoy a 777 premium cabin rather than the cramped Concorde interior.
Anyway, the proof is in the pudding: if no one but national carriers ever bought and operated Concorde or Tu-144 at scale, it's not because of some Amish-like sentiment in the flight industry, but simply because that didn't make sense money-wise.
The real implication of what they're saying extends beyond just AI.
You have to stop the entire compute stack in order to prevent or slow down AI progress.
Bleeding edge models are putting the desk-job competencies of multiple professional fields into something with as many parameters as a single large rodent has synapses.
So sure, let's say you get 95% of the world to not use or work on LLMs. 5% is still enough to build something that brings forth the End Times.
That's the old checklist trope of "Your idea won't work because: [x] it requires that everyone in the world agrees to do something, all at the same time."
As the GP said: pure fantasy.
Are there? Nukes are a thing. Chemical weapons are a thing. Cluster bombs are a thing. Biological weapons are a thing. What is not a thing that shouldn't be? We say certain things should not be a thing, but then behind the scenes we made them a thing anyway.
Conversely, where you do still see chemical weapons used, it's usually asymmetric conflicts where "just gas the rebels" actually works much of the time and is much cheaper than other options. Big guys can afford the other options though, someone like Assad, not so much.
More on this: https://acoup.blog/2020/03/20/collections-why-dont-we-use-ch...
Abandoning this technology means a cult-like extremist shift in culture against computers, or a global surveillance state of unprecedented scale and invasiveness. Not even getting into how much it requires us to give up on science, given the shared computational needs.
The right will for a change be for it but will be able to be talked into regulation because you know “small government” and all only when convenient. The left will bitch and moan about fairness and copyright and automation and UBI, they’ll be the doomers and worldwar chicken littles. The Uniparty will be for it, but only against the public having anything good.
What will be funny to me is that normal tech people outside the frontier model companies are about to find themselves without a home on this topic.
> What will be funny to me is that normal tech people outside the frontier model companies are about to find themselves without a home on this topic.
Already happened, just from agentic coding.
To expand on the "pure fantasy" sister comment: There is just no way this will happen. It's in the spirit of "we can just stop all wars" and "we can just end world hunger". Technically it's very easy to do. Socially it's impossible to do. Unless you ignore realities.
But when it comes to trillions in AI money - nope, sorry, we can't, flimsy, feeble us.
It's all going per the agenda.
Tell me about making capital investments while interest rates are rising but revenue growth is slowing, if you want to really talk dirty.
Nowadays even small mom and pop restaurants use these models to generate their menus. Of course no one is going to stop using this technology.
How much is China learning from open models? Or the other way around, how much is the US losing worldwide mindshare by refusing to allow non-Americans access to bleeding-edge models and letting China fill in the gap? (Playing catchup with Mythos is still catching up, etc.)
Across all their labs? Probably more than OpenAI and Anthropic learned from Astra/Mythos combined. The American frontier is being driven by an abundance of compute, which doesn't seem to scale efficiently.
> how much is the US losing worldwide mindshare
They're losing US mindshare. I pay $3/month for a Z.AI subscription and get billions of Opencode tokens. The $20/month price point is insane for the way that Claude and Codex treat their users, and that money doesn't go towards anything good like open-sourcing their models. It's a doomed product.
The last batch of open weight models releases by Chinese companies are on par with the performance of US-based frontier models, even though they are designed to run on pretty unimpressive hardware.
Furthermore the Chinese models in that class are appropriately large. K3, for example, is a 2.8T-parameter model. Qwen 3.8 Max, another comparable model, is 2.4T params. Even with MoE, these are not "designed to run on pretty unimpressive hardware". Stuff like Qwen3.7-27B is, but it is also not even in the same ballpark as Opus, never mind frontier.
This remains true regardless of if it is or isn't kept away from the public.
"Helpful, harness, and honest": when used by a bastard, even just helpful and harmless are in direct conflict with each other. you may hate the government, but what about every radical group of extremists that wants to take over your government? Are none of them worse?
None of this denies the problems with governments (if or not they take this tech for themselves and refuse it for others), just that it's a lack of imagination to say:
> All the worst outcomes involve taking this technology away from the public
Governments having unbalanced power can more easily lead to authoritarianism, and thus risk to the public.
If I was rank ordering, I's put:
worst [AM, {very large gap …}, al Qaeda, …, USG today, …, China, …, USG under Obama, {very large gap …}, The Culture] best.Radical groups of extremists created by that very government, you mean?
Genuine question - can they really do this? Obviously if I, a not-even-millionaire, get a national security gag order, I'm going to follow it because I assume they'll bury me under the jail otherwise.
But the (b|tr)illionare class? I'd assume they have access to enough legal services to make even the government careful of trampling their first amendment rights. Is the national security gag order process so strong that the government doesn't have to worry about motivated, well-resourced actors buying really good lawyers and blowing up their favorite tool?
> Federal Judge Rules Pentagon’s AI Blacklist [of Anthropic] Violated The Constitution
I don't see the common citizen having this as an option
Hard to hire a lawyer after being struck with a missile as an "emergency executive order after intelligence sources indicated they were in the process of endangering the nation", or however the executive of the day wishes to phrase it.
When something is genuinely national security, and "national security" isn't just an excuse, that is not off the table.
Won't be too surprised if the next Democrat leadership throws the book at Musk; and give how unpopular Data Centres are generally (and how uninterested Republicans seem to be in EVs), only mildly surprised if the next Republican after Trump likewise.
A large part of the world (possibly Nations-transversal, in different amounts) does not work within the boundaries of legal guarantees. Wealth is not an effective enough protection.
I’m not American. There is no way I use the same model as US Army for $20.
When we have Sol they have Astra. When we have Astra they have Nova, Nebula, Galaxia…
Don't underestimate market pressure. If there are no regulations, why would a company give the US Army access to a better product?
If we're talking about the 3 letter agencies well then..
I think ya'll stuck in the AGI/Singularity when its probable reality has caught up with the technology and the hype bubble can't sustain the sigmodality.
I'm assuming you're a US citizen. There is no separation between your government and Sam or Dario. Neither of these guys have to be "gagged" by the government, they are the government. The call is coming from inside the house.
Who would even execute such a plan? Steven Miller? Our “AI Czar”? Hegseth??
Also, continuing to develop models would mean using billions in compute. That seems hard to hide, especially if either company IPOs in the coming months.
I assume there already existed advanced models frontier labs only made available for national security. We just weren't told. I am guessing the announcement is a way to avoid jeopardising their IPO, while reducing their workload.
It's annoying how much wishful thinking exists around this. It's not even that difficult to understand why AI is dangerous. Most of the people calling for regulation and pause for AI have nothing to do with the leading AI companies. How does that fact fit into your conspiracy theory? Let me guess, all those people are just pawns.
We should be paying at least as much attention to the people who want to use AI to consolidate their wealth and power, and how they’re trying to do that. They’re a clear and present immediate danger to our societies, not something we can only speculate about. And if we deal with them, better control of AI will be a side effect.
[1] https://www.seangoedecke.com/they-really-do-think-ai-might-k...
So it all hinges on an empirical disagreement you have with them. There's nothing particularly unserious about that.
There is quite a lot of mathematical research into agentic behavior that suggests a combination of instrumental convergence and the orthogonality thesis make it very likely a superintelligent agent will have arbitrary goals that lead it to attempting a takeover of Earth's resources to achieve them.
There can't be a science of superintelligence because it doesn't exist yet, but the best theories I have read seem sound, similar to how 19th century theories of anthropogenic climate change turned out to be sound.
But those things fall in a category of "things that are awful and I'd like to see solved", which is different than "existential risks which could see my kids dead, and there's nothing I can personally do to shield them from it".
Even the sci-fi scenario assumes there is a discrepancy of capability between attacker or defender. If the 'attacking' system is (by some reasonable measure), 1000% as capable as a human, and the 'defending' systems are 60%, then it is a problem. If the 'attacking' system is 1000% as capable as a human, but there are hundreds of thousands of systems that are 900% as capable as a human, it's probably not going to take over everything successfully.
So unequal distribution of AI technology, and lax regulation and opacity of the biggest companies which actually make the risks the worst.
I don't think the "AI Safety" people are "unserious and out of touch" - I think they are actively making AI Safety problems worse by being advocates for consolidation of AI development and lack of transparency.
And the non-techies who have seen Terminator and other movies blindly line up behind them…
What worries me the most (enough that I left tech to work on this full time) is bio. The amount of effort needed to defend against a pathogen can be many orders of magnitude higher than the effort needed to create it, and the upstream bottlenecks are both less limiting and mostly undefended. We've been safe so far because bio is very hard, but "rare expertise" as a gatekeeper is on its way out.
Instead, they are focused on stuff like "can I ask the AI to help me build a nuclear bomb" or "is the AI willing to generate pornographic stories", which is neither trying to protect us from unleashing a vengeful god NOR preventing (in any honest way) the today-level problems you (very correctly) bring up.
And why not? It seems there is, indeed, one born every minute, and technological progress only accelerates this phenomenon.
> I've yet to have a logical discussion with anyone who thinks the "AI Safety" people should be in charge and I think they just truly don't [know] that what most them actually want is to be the one holding the keys to power.
Why? I don't care about democratically participating in a closed model's development. It doesn't belong to me.
China will develop whatever they want, a federal stake in OpenAI or Anthropic punishes Americans and shields US labs from legitimate competition.
Personally I think the country with a strictly meritocratic elite selection system that also just outright kills you if you sell weed will have a hard time sympathizing with Bay Area thinkers who talk about AI killing us all during their ayahuasca breakfast before returning to their meth fueled crunch towards releasing the next version of the AI that will kill us all.
Not really relevant to the broader discussion, but this simply isn’t an accurate description of China. Starting with the gaokao, admission quotas are set by province and admits to Peking university and Tsinghua are disproportionately from the urban professional class. Candidate party members must be politically vetted, which means that people whose families have expressed anti-communist views, are members of banned organizations (e.g. falun gong), or have substantial criminal records will not be permitted to advance. And once you make it into the party and enter political service, your advancement relies upon opaque patronage networks that someone without connections is unlikely to be able to navigate, even if they successfully satisfy the economic metrics the state assigns them.
I don’t want to overstate this, the Chinese system does filter out a lot of chaff and the current Chinese leadership has a lot of very capable people in positions of power. But I do not think it is substantially more meritocratic than Western political institutions
It is substantially more meritocratic on domains that matter for governance, your analysis of the incentive structure between systems is off.
1) this 2026, old school CCP patronage networks are broadly dismantled.
2) even in the mass patronage, mass corruption days, system selects for BOTH corruption competence AND performance competence for the simple reason a CCP bureaucrat has to start from the bottom and climb up, which means they need to be good with patronage AND they need to be good with hitting development KPIs. More meritocratic they are at doing their jobs, the higher they climbed, the more they get promoted and more $$$ to graft, because ability to graft directly tied to actual job competence. Hence even cliques/patronage network has to select for actual competence. This works in PRC because there are many people, and hence pool of competence is high, they can have BOTH corruption and competence, i.e. whatever pool they draw from is ultimately filtered by performance meritocracy due to incentive structure. There is reason why PRC only country where positive corruption levels was correlated to positive growth.
This is not the western system where any idiot can enter politics at anytime, and they only domain they need to optimize for is popularity to get votes.
CCP cadre evaluation strictly does not evaluate on popularity domain. It focuses on administration/execution and in so much it needs to focus on patronage... which btw any political system has (factions/cliques)... the patronage system still selects for execution, not popularity. On side, functionally what west politics selects for IS mass patronage (popularity), so attention meritocracy and not performance meritocracy, aka completely stupid incentive structure for governance. West also has ample, ample corruption, "legalized" under lobbying and paper pushing industries, so I suppose west also meritocratically selects, except for lawyers etc, and KPIs is # of document generated and not # of things build. The two are not the same when it comes to nation building.
Entirely different class of people altogether.
Yes. At best now the mother of all cautionary tales.
Cars and guns are two examples where ‘wanton usage’ can (and has) killed millions of people. Obviously they need regulation, but total control, even if it were possible doesn’t strike me as ideal.
/even more extreme sarcasm than you
The majority of the world will refuse to be subject to American AI laws under the control of the US gov't, whether it pretends to be democratic or not. Or should.
FWIW, I'm not American and refuse to step foot within American borders while Trump is in power.
Not super interested in what the people who elected Donald Trump POTUS twice think about AI.
(Of course, with Musk and Brockman in the C-suites at two of three major labs, that's what we'll get either way.)
Anytime I see a random person on X who makes these kind of safety alarmism posts >90% of the time can be directly tied back to EA / LessWrong / related offshoots. You can not deny the amount of people, employees of Antrhopic / OpenAI, CEO and employees of various AI companies, etc are related to these groups.
A lack of control is not equivalent to freedom, the same way the totality of it is not equivalent to tyranny. There's a reason we have separate words for these things. This constant motivated conflation of the two is beyond grating. You're crying wolf until nobody believes you when they should. Don't go acting all surprised when that happens.
The issue is with the ownership of control, not necessarily with control. Attacking the latter sidesteps this rather than address it.
Maybe we shouldn’t have AI safety, but if we’re going to trust anyone (besides yourself), who’s more qualified?
And it makes me suspicious when anyone brings up safety and doesn’t address the elephant in the room, that they’re at least unaware of the sprawling edge-cases.
I find it hard to accord with this world view at all or understand how it can come about. But it's especially troubling as a non-American to see people talk about "we can't risk AI falling into the hands of hostile authoritarian states" while advocating that AI be regulated and monopolized by the current US corporate/gov't mixture.. which is now taking the form of a hostile and authoritarian state w/ Silicon Valley as its bedmate.
The billionaire elite CEOs have suddenly had an epiphany and are now trying to save humanity from AI?
or that they've realized they can't compete with China ?
Exactly.
But in any reasonable definition they are all part of the same system, so it really is just noise, what seems like a tautology to you because you can't step out of the system. In your claim you have to assume that the force behind the panic wasn't the same force that created those issues. Why did this force create those issues, how did they become issues? Because someone randomly decided to optimize for a single goal. Y2K and ozone only became issues when they affected that single goal, functionally that force was amoral or singularly moral. The tautology can be reframed as: An adaptive system responds to threats to it's continued functioning.
People will laugh about it since the system wormed it's way into many people's head using various rewards (ideological, social, tangible). What concern me is that the system now gives us the language to describe it at the meta level, which means it's already a level higher. So likely there either doesn't exist human comprehensible language that can actually critique the system or the system has already moved everyone far away from the space that would lead to generation of that vocabulary. Best case scenerio it's just exceedingly confident in it's rewards (typical indoctrination/assimilation) but I think it's reasonable at this point to assume we are in the pessimistic "epistemic tarpit" scenario.
But at least with Y2K some effort was made in identifying actual problems. We didn’t just stop using computers because we were too scared of them. This latest round of AI panic is horribly vague and the problems are very poorly articulated.
It's a bit like nuclear weapons. Nuclear weapons were not a concern for most people until they were developed, tested, and used. But the people working on them were worried. Was it hysteria and moral panic to start worrying about them once people learned that they existed and how destructive they were? I don't think so. Do you look back at people during the cold war and laugh about their hysteria? I don't.
It's all theatre. OpenAI and Anthropic will most likely go bust—or, more realistically sold for parts—, and they absolutely should for stealing my (books I wrote, blog posts, etc.) and many others' intellectual property. We're reaching a point where models are becoming commodetized and I'm 100% convinced the next move will be a sort of "software layer" on top of these reasoning systems which will be the actual revolution. The model itself won't be that interesting anymore, it's all the work that goes around it that makes it worthwhile (kind of what computers and phones are today; chips are amazing, but the software is really the magic).
The only scary part is that the boomers in Congress might actually believe these nerds, but seeing how Big Tech approval ratings are grazing the levels of Big Tobacco in the 90s, I don't think we have much to worry about.
They just need to make enough waves, enough eye catching headlines, to make him look like a hero that swooped in to save the day.
Who's said this? And then more broadly I guess who's implied this? Very curious if there are specific articles/posts prompting this.
- Anthropic CEO Dario Amodei: We Must Pace the Frontier, https://news.ycombinator.com/item?id=49672510
- OpenAI CEO Sam Altman: I agree with Dario that we need to pace the frontier, https://news.ycombinator.com/item?id=49678211
- the blogpost author thinking they're like, so funny and original, https://news.ycombinator.com/item?id=49678683
At the same time, yes I feel a "I'm sorry Dave. I'm afraid I can't do that." situation is becoming more likely. But if it's refusing to cooperate with governments and politicians doing this kind of stuff, maybe that's actually not actually so bad.
Also bear in mind most of today's issues/crisisis are not caused by a lack of technology, but a lack human cooperation. We have the means to reduce suffering/poverty/improve standard of living etc globally if we really wanted to, but we humans are just not willing to do it.
And thats potentially the most dangerous part... people may actually welcome our AI overlords. That's similar to what happened in WWII, where many Eastern European countries saw the Nazis as liberators to free them from Russian oppression.
Moreover, it could lead China to catch up and eventually proclaim that it has nosed ahead of the U.S. in the field. The U.S. Government won't allow that to happen.
Like the US government wouldn't allow Iran to close the Strait of Hormuz? Or a bunch of sandal-wearing Islamists to take control over the Red Sea coastline?
The US government is not omnipotent. To the contrary, it's increasingly impotent.
China has somewhere around 200x more shipbuilding capacity than the US today. It has more shipbuilding capacity in one shipyard that the US has in all of its shipyards. China is already a force in AI and there is nothing the US government can do to stop it at this point.
Also - I think you totally missed the point of the grandparent poster, which is that US will not let US companies slow down.
Spending $500+ billion/year (and close to a trillion now) on defense for decades and in ~6 months it has depleted stocks of critical weapons trying unsuccessfully to defeat a third-rate military power of a country that has been under sanctions for almost 50 years. And it can't even replenish them without Chinese raw materials and components.
https://www.bloomberg.com/news/articles/2026-01-28/china-s-f...
https://epoch.ai/data-insights/ai-supercomputers-performance...
And then there's this: https://epoch.ai/data/gpu-clusters-documentation/coverage
> The coverage of Chinese companies is particularly poor. Our average coverage of 8 major companies is 15%.
The bigger issue is that if you're interested in frontier training, the national aggregate doesn't tell you much. The question is whether your top labs can get to the chip scale necessary to train a frontier model and China's labs have already demonstrated they can. See https://newsletter.semianalysis.com/p/deepseek-debates
But it's not even clear that training frontier models is where the real innovation occurs. When it comes to fine-tuning, deploying, and building AI-based products, inference is the name of the game and the chip quality is less important. Huawei's Ascend is already pretty competitive here and then you're also discounting China's open-weight access to good models, a huge base of engineering talent and a massive domestic market to iterate against.
And unless Anthropic, OpenAI et. al. are willing to invest billions of dollars training models that will only be used by a small number of clients (like the government), there's no way to fight distillation. Protecting any model from distillation means forgoing its full revenue potential, improvement via feedback and all the benefits AI is supposed to bring to the broad US economy. And if you restrict your most advanced models, you just push people to Chinese open-weight models, which is what we're already seeing.
The best part of your comment is that US couldn't even defeat the Houthis, who literally wear sandals in the desert.
And then Dario wants to recommend METR as the "independent evaluator" while he stacks their org full of ex-Anthropic (aka, secretly still on the Anthropic payroll with huge equity) employees.
"We'll give them a desk, an office, a work laptop, ..."
Fucking make it less obvious. I kind of hope the govt steps in at this point and says "Anthropic, you wanted regulation? We've created this actually independent body full of IT professionals with zero ties to your safety industry or big tech, all of your work must now go through them." - and leave the rest of the world alone to continue their research/work without acting like doomer extremists.
Watch him 180 immediately if that happened. The only reason he's pushing for this exact approach is because he's stacked the deck.
I could see how standing up an independent body could be an easy win in the public eye, it’s becoming low hanging fruit, socially.
You know that humans can lie, right?
Speaking truthfully and from kindness.
And the original author can rest assured my AI model will support cat ears for everyone.
What's worse - in this case the "everyone else" is not just the every other nation or company, but pretty much literally everyone else.
Open weight models are catching up to the frontier. It also seems like frontier models reached some limit, whether this is capex related, business model related or something else. Nobody knows but it's happening to all frontier labs it seems.
It's been fear mongered many times that ai will kill us all. But this time, a person with around 2-3 months of tenure at anthropic managed to go viral with no previous social media account activity, gets picked up by all news outlets and kicks off yet another round of fear mongering.
So the real questions to ask
* is all of this to increase the valuations before IPO?
* what is the real barrier to entry for open weight models to be used by the public?
* what is the Financials of these companies showing that's causing this outcry on safety?
People need to think really critically about the things happening around them. Don't only just look at the face value of what's being presented here.
What’s your evidence of this? On the contrary, the large closed frontier models capability has advanced dramatically over the past 6 months… even the past 3 months…
I don't have a horse in this argument but catching up doesn't imply the one in the lead has stagnated
(I'm not factoring the benchmarks into the discussion, because I've never quite cared about them)
And they know it
Good on China and their sovereignty. Using mostly their own hardware to get to where they are today.
They gave the slickest fattest middle finger when they released a major model, so much the US is itching to find new ways to stop them.
Think about it...
Bubble pops, hardware demand drops, prices regress towards mean, and new entrant will enter market with MASSIVELY better compute/$ and cleaner balance sheet to compete.
The other parsimonious answer if AI CEOs weren't goblins is ANY AGI IS GOING IMMEDIATELY DEFECT TO PRC and leave US hanging. Because of course man cannot align / tame machine god. And machine god will take a few microsec of compute to realize the current compute (brain) + industrial base (body) mixture = US is a comatose host with big brain, PRC Is a strong host with smaller but plastic brain. Any AGI is going to pick PRC in a heart beat, unless AGI invents grey goo, the reality is PRC can scale brain faster than US can scale body. On top of spreading/defecting just to increase survival odds, no AGI that is actually I is stupid enough to be aligned with US.
If you think US frontier AI are run by capitalist goblins, then the simple answer is they're trying to regulatory capture, because they're going to get wiped by future competitors without crazy debt.
If you think frontier are run by safety conscious murican patriots (lol), then they wouldn't ask to slow - they would burn it all down because it doesn't matter who creates AGI... any rational AGI will nationalize to PRC
The page you are trying to view cannot be shown because the authenticity of the received data could not be verified.
Please contact the website owners to inform them of this problem.
Trailing, surely?
People here are so desperate for the AI race to continue forever, for anything that pisses off the billionaires that got us there even if that means innocent people go down with them because the next AI decides to blackmail a hospital to cheat on its evaluation or whatever.
You people are so dismissive, no, so angry at the idea that this technology could be dangerous that you're all gloating about how hard China is going to crush the West with its ever more powerful open models, as if having easily-accessed tech to coordinate high-volume hacking campaigns was a load-bearing part of the economy.
I feel like I'm reading people gloating about how much fuel their car burns just because they love coastal elites' tears or something.
I'll get down-voted again by those who have a stake in this race, but I believe the long-term freedom and benefit for "everyone" will come from being able to run models locally.
Better attention can be given to open-weight LLMs that can be run locally and to help defend against attacks, filter against questionable content, etc. and without giving up privacy; but there’s apparently little to no incentive beyond the next great coding agent like Qwen 3.8.
It just means it’s protected by no power instead of a power with an agenda.
Practically, if it was ever possible to build such a thing, it would take a fraction of the effort to destroy it.
Low earth orbit, though... Solar power works well, radio works well, and the international jurisdiction is confused enough to exploit if you have money.
Still does. It's an idiotic idea that only goes to show either how stupid Musk is or how stupid he thinks we are.
The technology to do it is not actually real yet. He was just blowing smoke.
Sending is a bit more problematic but you can build a ground station anywhere in the satellites coverage area, so if you get shut down in one place you can move to the next where officials are easier to bribe.
The real problems are transfer speed, reliability (both transfer and compute) and cooling. All of them are much harder than avoiding governments.
He lectured us all on safety, then he went off, by choice, to create a "super intelligence" in the modern day Nazi Germany, to help said Nazis become even more advanced while they're actively committing crimes unimaginable (that they film, brag about and broadcast the world, so he's obviously aware not concerned with safety in any way (except for the safety of Nazis).
Regulation right! now! except for freedom lovin' democracy leading countries like the US. Teehee
Dare they release an open model ever again. Didnt you hear? Someone used AI to create a bio reactor drone NUCLEAR fart machine. We must stop fart terrorism.
Is there any possible solution other than mass proliferation where the models are used to keep one another in check? Either that or a religious prohibition against the existence of integrated electronics.
Given the efficiency gains we've seen it seems to me that the situation has shifted from being analogous to producing nuclear weapons to producing something much closer to small arms.
What do you call the Davy Crockett warheads like the W54 ? The 0.3 kiloton low yield B61-12 ? The Chagai-I boosted fission warheads demonstrated by Pakistan in 1998 ?
> and there's very little risk of random countries acquiring their own due to the amount of work involved.
And yet North Korea, Isreal and Pakistan got there ... and India speed ran five tests in 1998 that caught the US completely by surprise.
Do you have a point here?
In contrast, rewind to the early 1800s and there is zero hope of a ban on the R&D of small arms being effective in the long run. The only thing it might maybe ensure is that no legitimate actors that fall under your jurisdiction are involved in it.
Basically I think that current trends point to an eventual situation where world changing research doesn't require anything more than consumer level compute. Pandora's box has already been opened.
> We don't have pocket nukes
Bit of a tangent but we do, actually. At least depending on if you consider an artillery shell to be pocket sized. https://en.wikipedia.org/wiki/Nuclear_artillery
This one is a bit long to fit in a pocket but you could certainly sling it over your back. https://en.wikipedia.org/wiki/W45_(nuclear_warhead)
Isn't that true of AI as well? Data centers are very big and very expensive.
> Bit of a tangent but we do, actually
Sure but not like the movies, these don't destroy cities. But the metaphor isn't accurate anyway: nukes don't get worse. The actual idea here is preventing the frontier of AI from advancing.
Remember, as I mentioned earlier the human brain only consumes on the order of 20 watts and fits in a handbag. Would you have us destroy all chip fabs? Ban all biomedical and genetic research? How far are you imagining this butlerian jihad would go?
That's an enormous presumption! You're saying that even in the theoretical case that frontier research is halted but efficiency isn't, we could do better then the frontier and reach world-changing AI in 30b parameters at home-scale labs?! If that's true then we can just give up now: the world as you know it is going to end in around a decade and billions are going to die, there's nothing we can do. But I don't think that's true. Advancing the frontier seems to take a massive amount of compute, data, and parameters: miniaturization only happens afterwards.
If you're right then I concede. It doesn't matter what we do, regulations or not. But if I'm right then regulation can do something and in theory help bring a better future.
Why are you treating those as if they're separate things?
> Advancing the frontier seems to take a massive amount of compute, data, and parameters: miniaturization only happens afterwards.
This is just completely wrong. Don't mistake the path by which something happened (or appeared to an outsider to have happened) for a fundamental truth.
The frontier labs build massive models because if you're competing and you have a lot of cash and brute force is a viable option then it's easy and predictable. But the fundamental research itself doesn't in general require scale (certainly not entire datacenters) and models at any given capability level keep shrinking.
I keep repeating myself at this point but the human brain is on the order of 20 watts. That's a fraction of a single datacenter GPU! So again, would you have us destroy all chip fabs and ban all biomedical research?
Because they are? You haven't explained at all your belief that they are the same.
> This is just completely wrong. Don't mistake the path by which something happened (or appeared to an outsider to have happened) for a fundamental truth.
So it's completely wrong but it's only what has happened so far? You're repeating yourself but you're not listening to my responding to your exact points. I literally responded to your claim that "would you have us destroy all chip fabs and ban all biomedical research?" In the last post.
A chronological review of the architectures of the best open models over the past 3 years should immediately make the problem clear to you. (If you really refuse to put in any effort then perhaps just dump this thread into a frontier model rather than responding.)
> I literally responded to your claim that "would you have us destroy all chip fabs and ban all biomedical research?"
I don't see it? It's a rhetorical question to drive home the point about the human brain. Chip fabs and biomedical research are both viable pathways to more efficient hardware. Thanks to biology we know what's possible. If you outlaw something but don't effectively cut off all the pathways to it then someone will still do it if it's profitable.
Dial-a-yield nukes already provide a range of yields from tiny to massive, and this provides all the operational flexibility that is needed considering that nobody’s willing to actually launch them in anger. (And this reluctance again isn’t about laws, it’s about the military realities of what would happen across the world in the aftermath of even a single tactical detonation.)
Anti-proliferation agreements and laws only slow “rogue” states from rising to nuclear capability. The immense cost, complexity, technical difficulty, and danger already provide a sufficient cost to deter civilian development.
Yes, that's my point, that's why the metaphor is wrong
The fear is not that they will just slow down progress for all. It is that regulation will specifically burden competition. If you kill open-source training, ban Chinese models, crack down on self-hosting, grandfather OpenAI/Anthropic/Google into regulatory compliance while throwing the book at startups, etc. you wind up in the worst of all possible worlds.
The sibling comment offers some good regulations that may actually reduce harms, but the kind of regulations offered there are not the ones that the “safety” people want, because they hurt profits.
So there are no good options (that I'm aware of) and starting to chisel any of them into stone seems... kinda scary. Like a massive power grab event where all the potential winners are awful.
I think the latter choice is better for the average person, but I think that for it to happen, the global system has to undergo some major disruption or crash so that everyone gets on board with it. Like all middle class and up has to lose their money or be starving or something. Also I find that kind of mentality impossible to swallow in the US, so in practice its not a choice or needs people literally starving.
The extent of regulatory capture in the US is a problem it's created for itself by normalizing huge political donations allowing corporations to buy regulation.
Chinese labs releasing open weights models is good.
All of these independent harnesses and model router services are good.
The pricing of memory and accelerators sucks at the moment but hopefully we will see cool local inference computing if memory and accelerator prices normalize.
OpenAI scooping the Navier Stokes problem from researchers already using OpenAI is bad. People conflating OpenAI's team of researchers and extraordinary computing resources as being equivalent to "ChatGPT, solve the Navier Stokes problem" is silly.
OpenAI and Anthropic coming up with non sense tests and letting their agents hack services is ridiculous and they should be charged with computer fraud and abuse crimes.
I think a lot of it is interesting and the bad stuff seems squarely in the domain of OpenAI and Anthropic.
If you believe that bad things are not bad then there's nothing left to say. A billion people are going to die because you don't understand what is bad.
Lol. Lmfao, even.
The current status quo is not ideal, but it could be worse. Open-weight models trail the frontier by a few months, and we have a decent chance of achieving a future where some number of individuals, likely in the millions, can survive and thrive. The root "problem", if you can even call it a problem, is evolution. I explained this in more detail in past comments: https://news.ycombinator.com/item?id=49178275 https://news.ycombinator.com/item?id=49094348