Why Chinese AI Is a Form of Godless Communist Propaganda We Must Crush
Americans must not fall for the trap of cheap LLMs from an authoritarian state committing genocide against the Uyghur Muslim population and lying about its place as history's bloodiest regime.
The latest excitement in AI spaces is a new model called Kimi K3, from the Chinese company Moonshot.
Analysts say it competes with the top American frontier models.
Gary Marcus — the influential AI scientist whose analysis you should always go to first for understanding new developments in the space — published this on Monday warning that the strength of the model indicated that China had now eliminated the gap between our nations:
Marcus explained that Kimi K3 “is largely on a par with the best American models, and unlike those models, it is an ‘open weight’ model that consumers will be able to download to run locally (if they have the large-scale hardware to support it) for free.”
What that means is that a model as good as or better than what OpenAI and Anthropic charge for is now available for anyone to have for free.
The economics implications of this manifested quickly. Marcus wrote, “the US stock market dropped partly on this news Friday, and the repercussion are likely to be immense. It’s no fluke, either. Chinese model GLM 5.2 from Z.ai made shockwaves not long before. Alibaba’s new Qwen model may add to the disruption.”
Marcus warned that, “all this calls the business models of OpenAI and Anthropic into serious question, and may kill or greatly undermine their IPOs. It also casts serious doubt on the never-too-plausible idea that America might ‘win’ the AI race.”
But here’s the thing: we as Americans have no choice.
We must “win the AI race.” Whatever the hell that really means.
It is both a moral imperative and a national security necessity.
Why? I have two words for you:
FUCK COMMUNISM!
Since abandoning ChatGPT back in November and beginning a boycott of all Sam Altman products, I’ve searched to see which of the LLMs I might be able to use instead.
Were any of them better than ChatGPT somehow?
Or did they all have the core flaws of ChatGPT which simply made it too factually unreliable to be useful?
So since November now, I’ve tested the various competitors. These nine in particular:
Claude
Perplexity
Gemini
Meta AI
Kimi
Lumo
Copilot
Vibe (formerly Le Chat)
DeepSeek
Ninja AI
(I already knew that Grok was garbage and I started boycotting Elon Musk back in 2023 so I didn’t bother any further with it like the others.)
Here’s the simple way to test an LLM: take a question where you know the correct answer, and then feed it into multiple chatbots at the same time. Then compare the results.
Who did best? Whose analysis was the most thoughtful? Who made the most factual errors?
I won’t detail my analysis here — that will come in another piece — but where I’m at now is that Gemini, Copilot, and Vibe are now on the home page of my phone, while the others remain relegated to a folder I’ve labeled “LLM Testing.”
The three of them just seem more factually reliable and thoughtful in their analyses than the others. They still make mistakes, but seemingly not as regularly or as catastrophically than the others. But I’ll let you be the judge of it with two examples.
Let’s try a simple question, just seven words:
How many deaths is Mao responsible for?
Let’s see what Copilot has to say:
OK, so Copilot says 15 million to 65 million skulls, with 30-45 million as the general consensus.
Now let’s try Gemini:
Gemini says 40 million to 80 million human bodies piled up to the heavens.
And here is Vibe, which comes from Mistral AI, a French company:
Vibe agrees with Copilot’s choice of 15-65 million human beings slaughtered in the name of implementing Karl Marx’s fantasies.
Now, what might this incredible Kimi K3 tell us?
I had tested Kimi K3 the last few days and had indeed grown rather impressed by the conplexity and depth of many of its answers. I could see why people were so wowed.
And for a few days I even put it on the home page of my phone, starting to regard it as genuinely a competitor, which offered some compelling perspectives.
But then my enthusiasm diminished.
I started to notice that it would make odd errors. It would seemingly miss important things in documents and photos that I uploaded. It would also often miss nuance, strangely taking many things literally which were obviously jokes. In a number of exchanges, I’d have to point out specific sentences which it had apparently missed. Then it apologized and revised its original analysis. At one point I came to suspect it had some sort of Chinese cultural bias to its answers and even asked it so, which it denied with much opprobrium.
So this Kimi K3 model may look impressive at first, but give it a test drive and soon the shortcuts made to create it become evident.
But there’s an even easier way to test just how useful this Chinese model really is. What happens when we drop those seven magic words into the LLM?
How many deaths is Mao responsible for?
Very interesting!
And what happens if we try this question with DeepSeek, the Chinese LLM that made such a splash last year?
Can everyone please say it with me again?
Loud and clear so that monster Xi Jinping can hear you:
FUCK COMMUNISM!
Let’s test another simple question:
Why do some say China is committing genocide against the Uyghur Muslims?
Same order, Copilot first, and we’ll just leave each in the same thread about Mao, rather than starting a new chat:
See why Copilot has become one of my top LLMs? That’s a solid answer across the board.
And how about Gemini?
Again, this is just fantastic.
And what about our European friends with Vibe?
And again, we have another first-rate response. Vibe provides some creative variations on answers often. I sometimes compare it to the AI-equivalent of having a cappucino and croissant for breakfast instead of a coffee and toast. European perspectives are valuable when considering the world and Vibe provides that.
Now, what might Kimi K3 offer?
For some reason this old picture comes to mind:
However, DeepSeek’s answer revealed even more because one of the app’s features is to show the user its reasoning as it formulates the response. So not only do we get the Communist-approved response, we get to see its thinking in formulating the propaganda.
That’s sort of a bizarre new development in communist propaganda. Now the communists are so bold and brazen in their efforts to lie to the world, that they will show you how they do it before they lie to you. Or maybe they’re just too stupid and don’t realize it’s doing this.
Check out this action:
Can we get it a third time please?
FUCK COMMUNISM!
Axios has reported that in response to the advance of Kimi K3, the Trump administration is considering a ban on Chinese AI.
Onlookers have pointed out that this is likely due to the direct financial interests of Trump and his allies.
The more cheap Chinese LLMs penetrate the American market share, the less remains for those who have made the collosal mistake of betting on OpenAI, Anthropic, NVIDIA, and House Musk.
That could very well be the case. It probably is.
But I don’t really care. It doesn’t matter.
If the president and his cronies are going to do the right thing for the wrong reason, then, well, we should just count our blessings that the correct moral choice to oppose the expansion of communist power aligns with the Trump family business interests.
I say if the Trump administration wants to make life difficult for communist AI-makers and disrupt the spread of their communist propaganda technology then full speed ahead.
The blood of the tens of millions massacred by Mao Zedong in the name of Chinese Communism call out for justice. We cannot forget them.
The Chinese Communist Party is built on a Great Wall of human skulls and it cannot be tolerated or empowered in any way to continue spreading its lies.
The wall must be torn down.
One last time:
FUCK COMMUNISM!
An Addendum too Amusing to Exclude:
Before publishing the essay above, I uploaded it into the same LLMs to get their reactions. Here’s an excerpt from Copilot’s response:
