lossolo

Sure, these are my sources. World Bank (2023)...

My WTO comparison used the much broader "medical goods" category, so it was not an apples to apples.

My source was Citeline's 2026 annual review at https://pulseforinnovation.org/by-the-numbers-citelines-rd-a... which states...

EFPIA, using Citeline’s Pharma R&D Annual Review data, says that among the 104 new active substances launched for the first time on the world market in 2025, 46 came from Chinese headquartered companies, 28 from US companies and 16 from European companies.

https://www.citeline.com/en/rd26

https://www.efpia.eu/media/owqczcqz/the-pharmaceutical-indus...

China currently leads in this particular output measure of newly launched active substances.

Your clinical trials claim also remains unsupported as worded. Citeline’s figure of more than 11,600 medicines being advanced in the US refers to the drug development pipeline. It is not a count of active clinical trials being conducted in the United States. One medicine can involve multiple trials, and pipeline geography may be assigned through the developer's headquarters rather than the location of trial sites.

Citeline itself describes this as the number of drugs in the active R&D pipeline:

https://www.citeline.com/en/rd26

The most relevant peer reviewed international comparison I can find points in the opposite direction for new trial registrations. A 2025 study found that China registered 16,612 trials in 2023, including 7,798 randomized trials, compared with 9,100 trials and 4,619 randomized trials in the United States:

https://www.jclinepi.com/article/S0895-4356%2825%2900124-6/a...

That does not by itself prove that China has a larger stock of trials currently classified as active. It does show that "the US has more active clinical trials than any other country" requires a specific global dataset + status definition + date + deduplication method etc. A count taken only from ClinicalTrials.gov is not a neutral worldwide comparison, because US law requires many FDA regulated trials to be registered there, whereas studies outside its legal and policy scope may be submitted voluntarily:

https://clinicaltrials.gov/policy/fdaaa-801-final-rule

https://clinicaltrials.gov/about

For global comparisons, the WHO's ICTRP is more appropriate because it provides access to ongoing and completed trial records supplied by registries around the world and groups multiple records referring to the same trial:

https://www.who.int/tools/clinical-trials-registry-platform/...

IQVIA's latest result does support a narrower US leadership claim: trial starts became increasingly concentrated among US headquartered sponsors in 2025. But that measures the headquarters of the sponsor (not necessarily the country where the trials took place) and it does not measure the total number of currently active trials:

https://www.iqvia.com/insights/the-iqvia-institute/reports-a...

So the fair conclusion is that the US remains one of the world's largest clinical rial and drug development centres and may lead certain sponsor based or industry sponsored measures. The categorical claim that it has more active clinical trials than every other country has not been demonstrated by your sources.

In general, I was referring to "major industries" ....

I also do not think the UN classification resolves the "every major industry" question. ISIC is a statistical classification system, not a rule for what ordinary speakers must regard as a single competitive industry.

ISIC Division 30 combines:

shipbuilding railway equipment aircraft and spacecraft military vehicles motorcycles and bicycles

https://unstats.un.org/unsd/publication/seriesm/seriesm_4rev...

Under that aggregation, US aerospace strength can be used to declare the US a "leading player" in the division even though its commercial shipbuilding industry is negligible by global standards. Likewise, ISIC Division 27 combines batteries with motors, generators, wiring, lighting and domestic appliances.

Whenever the US is weak in one globally important industry, it can be bundled with another industry in which the US is strong. Almost any large, diversified economy could be described as a "leading player" in every sufficiently broad category using that method.

You are correct that the US has "only" 60 GW of domestic solar module production capacity...

The solar comparison also switches metrics. China having more than 80% of global solar module manufacturing capacity is a manufacturing claim. The US being second in solar electricity generation is a deployment/generation claim. Operating solar panels (many of which depend on an overwhelmingly Asian supply chain) does not establish leadership in manufacturing them. The IEA says China has over 80% of module capacity and 95% of wafer capacity.

https://www.iea.org/reports/advancing-clean-technology-manuf...

True but the US was the 2nd largest producer. In battery cells, we have an estimated capacity of 96 GWh in 2026....

The same issue applies to batteries. The 96 GWh number in your source is projected for 2026, not actual 2025 production. Specifically US energy storage cell capacity, not all lithium ion batteries and capacity, not output.

Your source says US ESS cell capacity was "essentially zero" in 2024 and projects 96 GWh in 2026.

https://poweralliance.org/2026/03/18/american-energy-storage...

Meanwhile, the IEA estimates that China manufactured well over 80% of all batteries in 2025. It says the US and EU each supplied a similar share of the relatively small remainder, while US and European factories remain heavily dependent on imported components. For grid storage LFP batteries specifically, supply is almost entirely Chinese.

https://www.iea.org/commentaries/global-battery-markets-are-...

Finance mostly true, space launch also true, cancer survival broadly true (with qualifications), oil and natural gas also true.

(exports nearly 2x the second largest exporter)

Can you provide the source? In the WTO's broader medical goods category, Germany actually exported slightly more than the US in 2022 ($202.6 billion versus $189.6 billion), so such a huge difference in a few years?

and pharmaceutical research

China accounted for approximately 44.2% of the 104 new molecules in 2025, compared with 26.9% for the U.S. and 15.4% for Europe.

The 2024 shares were approximately 34.6% for China, 30.9% for the U.S., and 22.2% for Europe.

there is literally not a single major industry where the US is not a leading player.

I will just give 3 examples: China completed roughly 91% of the world’s shipbuilding tonnage in 2025, holds more than 80% of solar module manufacturing capacity and produces more than three quarters of global batteries.

I don't really know how else to express my experiences living in, working with, and interacting with people in both of these countries.

Could you please describe your experiences, since you didn't? I've also been to China twice now (also multi month trips). I visited Tier 1-3 cities like Shenzhen, Shanghai, Beijing, and Chongqing etc, as well as some smaller cities. I'm very curious to hear about your experiences. How did they differ from the West for you?

Yeah, agree on both points. You can embed any bias you want using RL, regardless of training data.

These models are trained to be truthful

A more accurate statement would be that these models are trained to fit the training data as closely as possible, regardless of whether the training data reflects the truth.

It doesn't really address the point. Chinese students wanting to attend US universities is evidence that US universities remain attractive, not that those students would otherwise permanently immigrate to the US or that China lacks attractive careers for them afterward.

US immigration policy may be unnecessarily pushing away talent but the assumption that talented Chinese researchers would naturally remain in America unless prevented from doing so ignores the growth of Chinese universities/labs, companies, their funding, national prestige etc.

I mean, don't get me wrong, US is still highly attractive, it is just no longer the only place where an ambitious Chinese researcher can do important work and grow.

Your comment feels like an outdated brain drain model where talented Chinese researchers naturally want to leave China and the only question is whether the US lets them in.

That may have been closer to reality 10-20 years ago, China is a different country now, what I mean by that is they offer research funding, they have huge digital behemoths (alibaba, tencent, huawei, bytedance etc), large scale deployment opportunities and prestigious careers. Many graduates return because the opportunity set is attractive and they want to return, it's not just because US immigration policy pushed them out. Some also want to contribute to their own country's technological progress (which is a normal motivation btw), like probably you are also a patriot and want your country to succeed.

So, really, China's AI progress is not mainly the result of America failing to absorb every talented Chinese researcher. China has built a domestic ecosystem capable of producing and keeping top talent itself. I feel like a lot of Americans do not understand this.

Have you used the same session for audit? so switched to K3? or used a new session for K3? K3 is sensitive to this, they wrote about it on their blog.

If you can create a ship that is capable of 1g acceleration, it doesn't just stop accelerating as it reaches the speed of light

For any object with nonzero rest mass, reaching exactly the speed of light in vacuum would require infinite energy.

US ownership in capital markets by wealth level:

Wealthiest 1% of earners - 50.1%

Top 90-99% of earners - 37.3%

Top 50-89% of earners - 11.7%

Bottom 50% of earners - 1.1%

Source: Federal Reserve/Reuters

It is super difficult to mimic all of these signals in a way that would cluster the same as typical humans.

Not really, beat ML with ML. I won't disclose how to do it, because who knows who might read this, but you can easily do it with a model trained for that purpose.

you might be considering low-effort what’s actually an attempt at simplifying - which is arguably higher effort

I'm not saying that simplifying complex topics is low-effort, good simplification can obviously require a lot of work and I fully agree here.

What I meant is more that some of these tests feel methodologically sloppy, they are too shallow, miss important technical context, do not control for enough variables etc, yet the conclusions are sometimes presented lets just say... too strongly, as I don't want to be too harsh.

Yes, thank you.

He created Django, what do you mean he's not an engineer?

I specifically said that he is not an ML engineer (emphasis on ML), so I'm not sure what Python web frameworks have to do with anything.

Also 'low-effort??' his posts are extremely in-depth, clearly very thought through with a significant amount of time and energy

And yes, low effort. Pelican was low effort, his Fable test was low effort, his HN filter etc. Read the discussion in the comments under the Fable test, it's not just my opinion. There was also another example a few months ago. You can search for it, I don't keep track of these things.

I discussed this with him directly after he called himself an "ML expert" in comments.

This is a classic case of the Gell Mann amnesia effect. I read ML papers and work with ML, but to people outside the industry, his writing can look "extremely in-depth" even though it really isn't. People I work with have the same opinion.

clearly very thought through with a significant amount of time and energy. Additionally he does perform multifaceted checks across LLMs in many of his other blog posts.

I have never seen an article by him about any model that I would describe that way.

And the most revealing sign that he is not an expert is the type of questions he asks and the mistakes he sometimes makes in the comments here. They show why he is not capable of doing any technically in depth evaluation (at least with his current knowledge level).

If you actually want to learn something as a layperson, read articles written by ML PhDs like Sebastian Raschka or watch Stephen from Welch Labs etc. that are directed at general audience.

He is not an ML researcher or engineer, he is a passionate AI enthusiast blogger. He mostly does SVGs and other low effort checks (sometimes with major flaws, as people have pointed out a few times in the HN comments). Properly evaluating the model across all fronts requires a deep understanding of LLMs, how they work, the trade offs behind new architectures and the relevant research papers. It also takes a lot of time to build a proper evaluation framework so basically you can't just vibe code that if you want something that is solid.

Meanwhile, on June 12, two days after Anthropic sent the letter, the Commerce Department imposed controversial restrictions on Anthropic's latest Mythos and Fable AI models because officials feared they could be deployed by military intelligence users in China and other countries of concern.

So that was the real reason for the Fable restriction? Because Anthropic wrote a letter to the US government saying that China was distilling Fable?

Do we have any estimates on how much the price might increase? I was waiting for the MacBook Pro with the M6 Max and 128 GB of RAM, especially since there are rumors that it will come with a design refresh

When I was in China the only VPN that worked using China mobile and other local ISPs was Lets VPN, they route through HK.

If you exclude dividend paying stocks, then the entire stock market starts to look like a giant pyramid scheme casino.

Most SOTA humanoid robots use transformers, which are also at the core of LLMs. They also use NVIDIA GPUs for simulation training.

It doesn't for me. I use Fable to make plans, then give them to GPT 5.5 to review, and it always finds flaws and edge cases that Fable misses (some are really critical). It was the same with Opus 4.8. I'll admit it finds a bit fewer issues now, but Fable feels more like an incremental improvement than a major generation ahead.

Seems like free speech didn't help Americans avoid becoming economic slaves.

Probably. This is an 8-12 trillion-parameter model, which is why it costs so much, that is also a major reason, besides RL and synthetic data, why it suddenly gained these new capabilities. They claim it was not fine-tuned or trained specifically for cybersecurity, but is instead a general purpose model.

It would crash if not for massive public debt that went mostly into capital markets and huge inflation.

It looks like a budget car, not an exotic supercar.

now the EU, instead of making it easy for companies to innovate, spends billion on trying to catch up to the US. not even catching up. getting to where the US clouds are today.

What's your alternative? The US has behemoths with trillions of dollars in market cap, more than GDPs of most countries in EU. What kind of innovation in context of cloud do you think would allow anyone to compete with them? Who would risk their own money and pour billions into challenging them?

From the links you shared, it seems they are first trying to determine whether it's economically feasible at all.

No, they wouldn't, and they don't have it because they have chosen not to. There is something called an escalation ladder: you do not threaten to leave or kill your partner just because she spilled milk on your floor. That is the same reason Russia did not use nukes, and why other nuclear armed countries involved in conflicts have also avoided using them. The same logic applies here. Another example is that the US could bomb the Kharg island containing Iran's oil infrastructure, but that would be a major escalation. Iran would then have no reason to show restraint and could bomb the oil infrastructure of the Gulf states, creating a worldwide crisis.

xAI (part of SpaceX) using just 11 percent of GPUs[1]

1. https://wccftech.com/xai-using-just-11-percent-gpus-while-me...

Well, there are multiple token proposals processed in parallel, from which only one is picked, seems like branching to me. The only difference is that in case of CPU there is always only one possible branch that is correct.