linkregister

Do you get mad at your computer for replacing clerical workers? What does this nonsense comment have to do with the issue at hand?

What does your story have to do with Moonshot AI? Do you think they didn't also use the same corpus? Bizarre

You're calling someone a political hack, but imposing neutrality on my statement.

I don't even necessarily disagree with your assessment of this spokesperson. But you must admit how inconsistent you're being.

That's a reasonable viewpoint to have. In a multipolar world we want Mistrals as well as Deepseeks.

Your statement is orthogonal to my comment. Why reiterate the schadenfreude / fairness comment already stated several dozen times in this thread?

Reread my comment and look for a value judgement on my part. The final sentence is probably a good clue as to my opinion.

Commenters are overlooking the significance of this information and posting emotional reactions based on perceptions of fairness or feelings of schadenfreude.

The economic viability of Anthropic and OpenAI rely on their being able to charge more for model access than their R&D and inference costs. If the market price for SOTA model access drops below that level, then these businesses will have to decide whether to continue to lose money or to reduce spending on R&D.

Moonshot's papers [1] claim that their training load was primarily from synthetic data and model self-teaching rather than RLHF and therefore keep their costs low. If Moonshot genuinely does not rely on human-led training, they will surpass US closed-source model providers. The United States government considers US supremacy in "AI" as a national security consideration.

This announcement is noteworthy because it implies that Moonshot's success is in fact due to distillation. It's in the interest of US frontier labs to place barriers to this if they find themselves in the position of subsidizing rival labs' research.

1. Kimi K2, https://arxiv.org/html/2507.20534v1

It matters because the closed-source frontier labs spend lots of money on human data (RLHF / RLAIF with human oversight). Moonshot is accused of circumventing these costs. Frontier labs add research costs into their inference pricing. If the market doesn't permit them to sustain sufficient pricing to have a positive cash flow, then their business prospects become weaker and they risk insolvency. Furthermore, other leveraged companies are at risk.

The reason why the United States government is weighing in is because it's in the national interest of the US to have supremacy in "AI".

Legality or lack thereof is one of many data points about whether a thing is noteworthy.

Moonshot performing distillation is rational from their point of view. Reducing costs is in the interest of businesses. It's also rational for frontier labs and the US government to add obstacles to this process.

As consumers this is probably a positive development.

I think your characterization of my post, which credits Moonshot's innovation, goes a bit too far.

I'm happy that Kimi K3 is indeed SotA and its open weights are due to be released soon.

It's also true that Moonshot and other labs distill from Claude. This has been reported on extensively. I don't think there's any alpha for Anthropic distilling from this model. I do not mean to discount the tremendous amount of innovation regarding MoE and quantization that Moonshot has accomplished. But its training with synthetic data is in large part from distillation from frontier labs.

Everybody is amoral and cynically only doing good things for personal gain, except for you, the one good person

NOBUS exploits have rarely been a driving interest for elected officials. Trade restrictions and reciprocity are far more salient and legible. Most elected officials are only barely aware of what NOBUS exploits even mean.

Even during the pre-Snowden heyday of US cyber supremacy, these capabilities were barely part of the thought process of White House officials.

Can you refine that search term a bit more? I followed the link but it brings up many marginally related comments.

uses rhetoric

counterparty uses rhetoric in response

"Hey! No fair!"

The parent poster is almost certainly talking about inference within workflows and not for interactive coding agents.

It's almost trivial to create a custom Slack application wrapping your desired harness running in a container on your organization's k8s cluster. Likewise with MCP support. These are already open.

Is there an example of this happening? Isn't the problem that datacenters are drawing electricity at the market rate and driving the cost up, rather than paying a surcharge for the difference?

If you look at the annual tuition rates for public schools in the United States, you will find the UC system to be an outlier. I gave a system in the same state that is closer to the national median. How much simpler can this be communicated?

Truly torturing the data to index on student body size as the relevant metric

You could say the same about all government programs, including pensions and defense. Yet somehow popularly-elected governments keep finding themselves maximizing public utility. A curious property indeed.

The UC system is an outlier in both price and quality for public education in the United States.

The California State system, which costs $7-9 thousand per year, is far more representative.

Loses argument, insults commenter

Distracts from the much larger environmental impacts of coal, oil, and nat gas production, which is much higher on a per-capita basis.

Comparisons should be made for replacements.

Modern pretraining also consists of expensive human-led specialized task creation and grading loops. Synthetic generation and distillation from previous models is another input for training. I wonder how much new text contributes beyond keeping knowledge up-to-date.

Unit economics for renewables coupled with storage are excellent. I agree we should reform nuclear regulation to allow new nuclear plants to pencil out. I disagree that we should discount the value of renewables.

Open models enforce a floor on prices, unless overall compute is so constrained that those prices rise also.

Your argument assumes premises that are hardly certain. We use automobiles to travel, yet tens of thousands can still ride horses today. Hell, there are thousands of blacksmiths and glass blowers. If there's a several hundred million dollar market for programming language experts, this will provide sufficient incentive for some people to remain sharp.

We still have COBOL programmers for a reason. The economic incentive to keep the skill never left.

====

linkregister: LLMs are trained with a large corpus of commercially available source code. However...

This concedes your point. Hence the following "however". It's bizarre to argue against it.

LLMs are trained with a large corpus of commercially available source code. However, intentional training on individual skills such as medicine, finance, mathematics, and software engineering is conducted to the tune of a few billion dollars per year.

Your scenario would only unfold if frontier labs decided not to compete on capabilities. It sounds unlikely.

It's been heavily edited to reduce its AI-ness. Some tells remain, for example "Not X. It's Y."

Enough smell has been removed where it's tolerable.

In the US, attaching legislation like that to firearms instructions would doom its success. It sounds like a great idea to ratfuck a potential bill. No Republican would be able to pass a bill that bans local agents from helping citizens exercise their 2nd Amendment rights.

The politics are probably different in other countries. We're still seeing Chat Control efforts in the EU.