Yesterday U.S. vice-president JD Vance appeared at an Artificial Intelligence summit in Paris to push for the exclusive adoption of closed AI models which the U.S. attempts to monopolize.

(Article republished from MoonOfAlabama.org)

Vance is an excellent speaker and his sermon was well written. Its presentation can be seen here.

Arnauld Bertrand summarizes:

If you weren’t convinced that the US’s vision for AI is that of a geopolitical tool that it fully intends to weaponize, watch Vance’s speech in Paris’s AI Summit.He even literally says it: AI is a “weapon that’s dangerous in the wrong hands, but is an incredible tool for liberty and prosperity in the right hands.”

Who’s hands are the right ones?

The rest of the speech is incredibly threatening and actually quite dark:

• He says he’ll ensure that American AI dominates (“the United States of America is the leader in AI, and our administration plans to keep it that way”)

Since a Chinese company developed DeepSeek and published it as an open source model that statement is no longer correct. The better, cheaper and more open models that will dominate.

• To do so he says that the US will keep restricting access to “all components across the full AI stack” to “ensure that the most powerful AI systems are built in the US”. He also says that he will “close pathways to adversaries attaining AI capabilities” on par with the US

U.S. AI companies have used brute force methods to expand the capabilities of their (large language) models. The Chinese company behind DeepSeek used brainpower to develop AI algorithms which achieve better results with less resources. There is no way to restrict that. It is quite obvious which of those two methods will in the end win. Attempts to close pathways in science and technologies have historically failed again and again.

• He told the assembly that it’d be “a terrible mistake for your own countries” if they “tightened the screws on US tech companies”, which is as clear a threat as can be
• He essentially tells the assembly, made of many of the world’s leaders, to screw their efforts at taking a collaborative multilateral approach to AI (and the US was the only country, with the UK, that didn’t sign the closing declaration at the summit)
• He tells the crowd they’ll need U.S. energy to power AI and derides their efforts at trying to be self-reliant for energy
• He says that “AI must remain free from ideological bias”, by which he means that it must spread an ideology that he is comfortable with (and all the others are “biased”). You can see that by the fact that the example he uses for an AI that’s not used in an ideologically correct way is if it’s used to make George Washington black.

If you believe that U.S. AI models are free from ideological biases please go and ask an OpenAI interface about the Middle East conflict and the role of the Zionist colonial entity plays in it. The answer will be some mushy nonsense with little relation to historic realities.

A Chinese hosted DeepSeek model will likewise avoid answering questions about the 1989 event around Tiananmen Square. DeepSeek however is an open source model which can be replicated without deploying its Chinese censorship layer. Peter Lee has tried it and checked the results:

The most cynical effort to counter DeepSeek was the resurrection of that reliable China-bashing perennial, Tiananmen! To discredit the PRC upstart.As in, DeepSeek would not return results for a tasking on the Tiananmen massacre. Shame!

Well, shame on the New York Times and Bloomberg and the Wall Street Journal and The Guardian for running with this canard—a sign, I think, that the order had come on high to deploy the Tiananmen wunderwaffe in order to blunt the appeal of this big Chinese soft power and tech power win.

It’s a canard because, as the G7’s premier news outlets undoubtedly knew well, only queries made to DeepSeek’s own server in China would yield this outcome. With DeepSeek AI installed around the globe, DeepSeek outside China will happily regurgitate the sad story of Tiananmen in 1989.

In the transcript you’ll find a screenshot of the result DeepSeek R1 delivered for my query on the Tiananmen Massacre via a US based service, Perplexity. It’s got descriptions, casualty numbers, even the precious Tank Man is there!

Vance, Musk and the like are still trying to convince the world that their closed and monopolized AI models, which include a U.S. censorship layer, will be the only viable path.

The competition however will come down to usefulness and price. It is there where the Chinese models are destined to win (machine translation):

The United States continues to cling to the thinking of “small courtyard with high walls”. Closed source is the mainstream thinking. On the one hand, it controls the development direction and speed of artificial intelligence, and on the other hand, it monopolizes the economic benefits of artificial intelligence.But with open source and low-cost alternatives, the “high wall of the courtyard” may become a dead end, at least a branch road.

Since the best open source technologies come from China, the U.S. development community will build their systems based on these technologies and become part of the Chinese-led artificial intelligence ecosystem. The activity and inclusiveness of open source will further expand the influence of the artificial intelligence ecosystem, making China the center of new technology in the world – this is a huge danger to American hegemony.

Open source AI is indeed a danger to U.S. hegemony in the AI field which is why Vance is barking up the tree. Well, he can do so as long as he likes. He may even impress some dimwit pols in Europe.

But in a free global market the more efficient solution will most likely win.

Read more at: MoonOfAlabama.org