D

I love AI, AI loves me.

[D @quantduck]

世界可以分成好人坏人阵营吗?

[Will Manidis @WillManidis]

My friend Dean Ball has advanced an argument for the de facto protection of American frontier intelligence providers.

Dean does not propose banning Chinese open-weight models. Banning things requires Congress. He proposes something more characteristic of the modern administrative state: every agency issues enough warnings, bulletins, and speculative security notices that no regulated company will risk touching them.

Even a reader sympathetic to Dean would call this protectionism, and protectionism has a long history in America.

More precisely, it's a proposal to use the informal, coercive power of the terminal, late-stage bureaucratic state to clear the American market of a cheaper frontier competitor to OpenAI or Anthropic.

But throughout the history of American industrial protectionism, it has always had two features. First, it's done in the daylight, and two, it comes with a bill.

In the spring of 1952, the United States was fighting a war in Korea. Truman concluded that a shutdown would endanger soldiers abroad and ordered the Secretary of Commerce to seize and operate most of the nation's steel mills.

The Supreme Court sent him straight back to Congress in the Youngstown Steel case.

Justice Black, writing the majority's opinion, begins with the rule that Dean's proposal is seemingly designed to evade: that presidential power "must stem either from an act of Congress or from the Constitution itself."

It's easy to flatten the Youngstown decision into the proposition that the president could not seize a steel mill. Its actual lesson is subtler: that an emergency does not dissolve the difference between making a law and executing one, that the importance of the object does not create the authority, that the inconvenience of the regulatory process is not inherently a source of presidential power.

Truman's approach failed not because steel was unimportant, but because it was so important that the constitutional bargain had to be made and the policy had to be carried through the front door.

Much like policy proposals from the rest of the AI agenda, Dean is proposing a smaller action in formal appearance and a much larger one in practical effect.

We will not ban Kimi, we will not prohibit it from use, and we will certainly not publish a rule declaring Chinese weights unlawful. But we will whisper about it. A regulator may even ask management whether it has considered the reputational consequences of relying on the Chinese model, but the agency certainly will never be coherent enough to ask anyone to stop. It merely ensures that continuing becomes professionally indefensible.

This is how we grow the administrative state, with bureaucrats that we placed in these roles, without accepting responsibility for the actual process of governing.

America has tried this experiment before. Operation Chokepoint didn't make payday lending, firearm sales, or any of the other seemingly distasteful businesses caught in its net illegal, but it encouraged banks to understand that serving legally disfavored customers would invite regulatory interest. We didn't pass a law, we simply just asked, "Are you sure you really want to be doing this?"

Reputational risk was powerful precisely because it's not law. It has no limiting content. A regulator did not need to identify a violation or even a material financial risk. He only needed to make the bank afraid of being asked what was actually going on here.

The analogy is almost embarrassingly exact to Dean's policy proposal. Dean need not prove that a Chinese model contains a backdoor, nor prove that it uses any more distillation than American models do. He simply needs to announce that there may be one. The agency does not need to order a company to stop using it, but simply ask whether management has considered the risk. The absence of formal policy is by design.

The Supreme Court dealt with this technique in NRA v. Vullo. New York's financial regulator could not directly punish the NRA's speech, so she allegedly pressured the insurers and banks she regulated to sever their relationships with it. The Court's rule was unanimous: government officials may not use their offices to "coerce private parties" into suppressing what the government disfavors. The communication must be understood in the context of the regulator's power, including the regulated party's knowledge that the person offering advice can also investigate, prosecute, fine, and settle.

The current administration has gone even further. In April 2026 the FDIC and OCC issued a final rule to prohibit regulators from criticizing institutions, formally or informally, on the basis of reputational risk, and from encouraging banks to deny services to lawful but politically disfavored businesses. In June, the federal banking agencies removed the remaining references to reputational risk from their supervisory materials.

Dean is proposing that this administration recreate for AI the same machinery that all of us argued against when we were widely debanked.

A government that can quietly remove Kimi from the market can also quietly remove gun makers, crypto companies, churches, newspapers, or American open-weight models from it. The bureaucracy does not remain attached to the intentions of those who staff it at the current moment. You don't get to build this machine just because your friends happen to be in office right now and keep it pointed at where you left it.

Protectionism through a whisper is not a more modest protectionism than by law.

Protectionism also has always come with a bill.

OpenAI and Anthropic increasingly speak of themselves as national institutions. Their compute is "strategic infrastructure," their losses are "national security losses." Their competitors are not just competitors, but instruments of hostile states, and their access to power, chips, capital, copyrighted material, and public customers is a matter of national survival and great power competition.

When Washington decided that the atom was too dangerous and too important to remain an ordinary private business, Congress created the Atomic Energy Commission and transferred the Manhattan Project assets and responsibilities to it. Production facilities and reactors were government-owned, and technical information sat under federal control, and private participation only returned later through a statutory licensing regime. The existential framing of the atom by its greatest proponents produced public control.

When national security concerns helped to preserve AT&T's integrated position, that is, a monopoly, in 1956, Bell did not receive this protection for nothing. The consent decree required compulsory licensing of roughly 9,000 patents and restricted Western Electric's commercial activity outside the telephone system. The settlement diffused the inventions accumulated inside the protected monopoly into the broader economy before breaking it up just a few decades later.

The pattern is really simple. It's not that every tariff necessarily demands nationalization. It's that the bigger the shield you are asking for, the bigger the bill you owe to the American taxpayer. And OpenAI and Anthropic have been unambiguous about asking for the biggest shields of all time.

Listen to what they are asking for: public infrastructure, privileged energy, federal preemption of state law, favorable copyright treatment, government contracts, export controls, and a domestic market swept clear of their strongest price competitor, all filed under national security interests.

And what do they want to pay? Almost nothing. OpenAI has floated giving 5% of the company to the American taxpayer.

They would like the benefits of nationalization at the price of being an ordinary public company.

There is also a profound moral hazard buried in Dean's proposal, as well as adjacent commentary on this. The labs say the Chinese companies distilled their models. Perhaps they did. Perhaps distillation matters. And perhaps the Chinese labs are running distillation attacks on scales that the Western labs are. I can't be sure of this. But if the reward for failing to secure an API is that the government removes the resulting competitor, the taxpayer is paying the lab to be careless.

We know how to secure an API. Know-your-customer laws exist. Access controls exist. Extraction detection exists. If you spend some fraction of the hundreds of billions being raised to defend the asset whose theft is said to threaten the republic, you might be able to stop some of this.

Theft remains theft when the lock is bad, but the owner of a badly secured store does not receive ownership of the street for his failure to protect it.

Dean's fourth point is that open-weight AI ends in communism: the state builds the training runs and subsidizes the product of intelligence and gives the models away. But, at least for me, this is not a particularly Chinese idea, but one of the most American ones imaginable.

The roads we build are public. Our radio spectrum is publicly allocated. The government funded the early internet and much of the research base behind modern computing. The state is welcome to build a platform, and American businesses are welcome to be built on top. Just because they're bad for our market position doesn't mean we get to call them Chinese in some fundamental way.

There will be inference companies and application companies and security companies and fine-tuning companies and data companies and chip companies and 10,000 businesses we don't even have names for yet. A public road existing does not abolish the trucking industry, nor does it nationalize it.

Sure, this may reduce the value of a couple trillion dollars of equity in the first generation of model companies, but it's certainly not communism. This technology may be civilizational without its present owners being permanent. And that is the thing that I feel like none of you will say out loud: that AI is welcome to be a civilizational technology when we ask for support, and an ordinary private product when anyone asks what the public receives in return.

The United States has two honest options.

First, treat AI as a competitive industry. Then the answer to Kimi is a better model, run cheaper and exported harder, with written rules excluding Chinese systems from defense, intelligence, and critical infrastructure when a concrete security case can be made.

Or two, decide frontier AI is too important for ordinary competition. Protect the labs through pseudo-nationalization, guarantee there's a market for them, and exclude the rivals.

But in that second case, the American taxpayer must be paid, likely through a majority of equity in these companies, if not full nationalization.

What no one gets is that private upside, public infrastructure, government-mandated scarcity, and immunity from cheaper competition delivered through a late bureaucratic state issuing warnings is a disgusting ask for something that is easy to name: regulatory capture.

There is a serious American argument for protecting industries that we can't afford to lose. But there has never been a serious argument for doing it invisibly, for free, through a bureaucracy instructed to manufacture fear, even if we can do it because our friends happen to be in office right now.

If the labs want to be protected, they should ask for it in the way that Americans have always asked for it.

In public. With a price.

[D @quantduck]

如果垄断利润能加速创新，波音早就该把 SpaceX 甩在身后了。

[D @quantduck]

恰恰相反，AI真正的加速，不是少数公司垄断最强模型，而是让智能像拖拉机和互联网一样普及，提升每个人的生产力。

[Dean W. Ball @deanwball]

Some observations on Kimi:

1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run.

2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China.

3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex.

4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business.

5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this.

6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked.

[D @quantduck]

我落后了,我们投票禁止你吧🤡 操你妈!

[Demis Hassabis @demishassabis]

https://t.co/PTeDiv1b6L

[D @quantduck]

深表认同

[Satya Nadella @satyanadella]

https://t.co/xv6csf1SbV

[D @quantduck]

滥用 /goal 只会生成冗余代码，大模型臆想出的大多是伪需求。

[D @quantduck]

深表认同,哪怕遭到OpenAI灵魂人物伊利亚・苏茨克维发起的政变，他依旧稳住了局面。

[shafin @_shafinsiddique]

sam altman is probably the greatest ceo of our time. he practically has every major tech CEO investing hundreds of billions of dollars to compete with him, sometimes even colluding (see elon + zuck). I don't know if gates, zuck, or even elon had this level of competition when they were forming their companies. OpenAI still consistently churns out some of the best models and still emerges as the winner in almost every category. my only question is what did @paulg see during that 10 min YC interview

[D @quantduck]

最近，我尝试一句话表达，固定的句式让我语言更精炼，思维也更清晰了

[D @quantduck]

看来，我这周二十亿 Token的开销和团队平均差不多，AI成本普遍变高，大家都会少用了。

[dax @thdxr]

our team's first week of having both sol and fable
the crazy thing is 30% of our cost was on fable

[D @quantduck]

原本对 Fable5 抱有偏见，实测后发现它能引导我理清项目难题，而相同提示词下 GPT 的回复却十分机械直白

[D @quantduck]

好像也突然明白为什么马斯克要限制员工使用 ai 的费用了，如果不计成本的外包认知并不能提高项目的进度。

[Chamath Palihapitiya @chamath]

8090 works on production systems for large, often regulated, enterprises.

Vibing isn’t tolerated because these are the systems that run western society - banking, power, healthcare, insurance etc.

Over the last few quarters, the gains that we got from using frontier models inside of our Software Factory on these systems started to shrink but the costs kept doubling. This makes sense I guess, as in hindsight, we were initially asking the model to do mostly light work (generate basic PRs) and now we were asking it to do more complex work (mitigate dependencies across systems).

Unless you grow context massively, be willing to run many A/B tests and iterate massively (ie use massively more tokens) complex tasks stay roughly unfinished by the model and requires the engineer to largely act alone.

In other words, we find the last 5% (ie where a model is truly equivalent to a reasonable engineer) extremely difficult to achieve and extremely expensive to such a degree that the fully loaded cost of the model + the engineer will not pay for itself.

So I asked our CTO to start thinking about other ways. We need our engineers to have access to the best tools BUT we also need to educate them to think even more for themselves - not less - in this last mile.

At the same time, we need to find solutions that decrease our token costs by 90% - especially because these bleeding edge tokens are not nearly as cost effective as the tokens before it and are creating a big OpEx bill for us.

I wonder how many engineers, in all orgs, are running amok right now by using the latest frontier models as a kind of slot machine. Increasingly turning their mind off, largely keeping productivity flat while their CEO and CFO deals with a massive token bill?

My advice to you is that when you encounter this last 5% of very hard technical challenges in getting a complex system into production, be circumspect.

The challenge of the last 5% is actually getting harder - especially as hundreds and thousands of code generation model runs run amok adding all kinds of random cruft into codebases that eventually need to be rationalized.

[D @quantduck]

代码也是

[Ray Dalio @RayDalio]

All that I've said thus far will be useless if you don't have good governance. Governance is the oversight system that removes the people and the processes if they aren't working well. It is the process that checks and balances power to assure that the principles and interests of the community as a whole are always placed above the interests and power of any individual or faction. Because power will rule, power must be put in the hands of capable people in key roles who have the right values, do their jobs well, and will check and balance the power of others. #principleoftheday

[D @quantduck]

gpt-5.6-sol 今天烧了6亿token

[D @quantduck]

He doesn’t have principles. He just translates his self-interest into moral language.

[Claude @claudeai]

There’s hope in hard questions.

[D @quantduck]

真他妈小丑 🤡 如果真的是一个道德高尚的人会玩这些?

[Claude @claudeai]

There’s hope in hard questions.

[D @quantduck]

@steipete 早期竞争中只有邪恶的公司才会领先,因为它们没有任何约束. Anthropic真他妈小丑🤡公司

[D @quantduck]

而且还是一个🤡公司,泡沫吹太大了.
糊弄谁呢?

[D @quantduck]

我一直觉得Anthropic是一个邪恶的公司,爆出惊天丑闻是早晚的事情.

[Ravid Shwartz Ziv @ziv_ravid]

The problem with Anthropic's consciousness paper

My last post got more attention than I expected, and the question I keep getting is some version of "okay, so what is actually wrong with the paper?". Let me try to explain.

First, the core result is fine. Reading out intermediate-layer representations and asking which ones the model can actually use downstream is a real question, and people have been poking at it for years. The J-lens is a reasonable tool. If you strip the paper down to the linear algebra, it is a decent piece of interpretability work.

My problem starts one level up. This did not need to be a paper about consciousness. It did not need global workspace theory, it did not need the brain, and it did not need the word "conscious" anywhere near it. The same experiments, the same figures, the same tool, all survive perfectly well as plain interpretability. Someone chose to wrap it in neuroscience. That choice is the product, not the science.

At the end, this is the main thing. Almost everything Anthropic ships as blogs/papers/posts is PR. They build genuinely good models and they are even better at packaging them.

A publication used to carry a specific kind of weight. People spent years on something and wanted to tell the world what they found. It happened mostly in academia, with a few industrial labs as the exception, Bell Labs, IBM and Google (for a while), but the distance between the paper and the product was real. When you read a paper you could assume the authors were not trying to sell you something underneath the ideas. There were outliers, but they were the minority, and the researchers you trusted would not risk their name on a narrative.

We are not in that world anymore. Every startup now publishes blogs and papers to raise its visibility, and that is fine, that is marketing and everyone knows it. Anthropic does something more effective. They erase the line between legitimate research and PR. We get confused because the models are so good, so we assume the outputs are research. A lot of the time they are selling us something. Sometimes it is "our models are safer," sometimes it is "our models are more capable," sometimes it is positioning for regulation. The consciousness framing serves a narrative they already committed to, models that look more and more like the brain, from a lab that has publicly tied itself to AI welfare and moral patienthood. The direction of the push is not subtle.

If you want the sharper version of the technical objection (disclaimer: I'm not an expert) Global workspace theory is a theory of access, not experience. Ned Block's distinction between access consciousness and phenomenal consciousness exists precisely to block the inference this framing invites. Access tells you nothing about whether there is anything it is like to be the system. The paper is careful enough to say it demonstrates no subjective experience. But that disclaimer is not what propagates. What propagates is "consciousness" in the same sentence as "Claude," published by Anthropic, borrowing the vocabulary of neuroscience to lend biological weight to a subspace of activations. The paper keeps the rigor of Block's vocabulary and drops the rigor of his argument. Most people who see the headline will never read either one.

Of course a lab named Anthropic is going to anthropomorphize its models. But we should be able to separate a good interpretability tool from the story it is dressed in.

So that is my problem. Not the math. The narrative bolted onto the math, and our willingness to keep calling it research.

[D @quantduck]

一直感觉AI智力化不大,因为芯片算力并没有增加多少.
就像你本地部署的模型永远都感觉差不多.只是更擅长某些事情罢了.

[D @quantduck]

循环工程API比CLI更适合AI.