The Veep Speaks AI, but to whom?

Part 1 on Vance's European Adventure

When a VP speaks, it has the implicit message that this is what the president would say if the president were there. There are two important questions about a vice-presidential address: what is the VP saying but also, who is the intended audience?

With VP Vance making his first vice presidential trip abroad, delivering addresses to the AI Now Summit in Paris and the Munich Security Conference. I studied this trip and wrote about it all Presidents’ Day weekend—so you don’t have to. (Gentle reader, you can thank me by liking my posts in Substack!)

Today we’ll take a look at Vance’s speech in Paris. AI is clearly a priority for the administration, but also for me. Besides my interest in all things presidential, I’ve done a bit of work on AI policy.

The Veep as AI Sales Rep

At the Paris Summit, Vance’s core theme was that AI is transformational, the U.S. will lead the way with a permissive regulatory environment, and the E.U. risks falling behind because of too much regulation. (Here’s a positive take and here’s a negative take that includes a transcript of the speech.)

This core message is not unreasonable. The trope that Europe regulates and the U.S. innovates is not new. It isn’t entirely true. For example, the U.S. has firmer regulation of pharmaceuticals than the E.U. At the same time the E.U. seems better able to undertake massive urban infrastructure projects, while U.S. efforts are stymied by… regulation. (I wrote that a VP-lead task force could help get American building again but we went with DOGE.)

Regulation is not necessarily a barrier to innovation. A solid regulatory environment can enable innovation by enforcing standards and building public trust. But getting the balance between useful and harmful regulation is hard. Criticizing European trepidation about AI has merit, but European caution is not unreasonable given Silicon Valley’s “move fast and break things” ethos. There is a lot of AI snake oil out there that isn’t properly tested, is used improperly, and is opaque in its functions. Ideally there would be a lively push-pull between the U.S. and the E.U. that would bring us better AI.

In his criticism of AI regulation, Vance also made the important point: when the big players in a field call for regulation, it is to lock in their dominance.¹ This is true, but at the same time, good regulation can ensure that big players don’t lock in their dominance. This regulation stuff is hard.

The core assumption in Vance’s speech is that AI is transformational. That may not be true. He compares it to the development of the steam engine (which was at the very center of the industrial revolution.) I’m not a Luddite or anti-AI, I’m just skeptical about these vast world changing benefits. If it were my speech, I might share some specifics about amazing things AI has done which have had real tangible benefits… but that’s me, and I’ve never been elected to anything.

The Veep Making a sales call in Paris.

Vance’s speech was basically shilling for American AI (Vance was briefly a tech entrepreneur after law school and being a best-selling author), which highlights who his real audiences were. Of course, he was speaking to the E.U. and effectively telling them they are welcome to join the U.S. in AI development, but they need to reduce their regulations and improve their power grid to meet the demands of AI. He offered a partnership, but with no real specifics. His offer was: our AI is the best,² and if you do these things we’ll let you in on it. Not sure if there was much on offer to attract the Europeans.

The Real Audience

The U.S. economy (for the moment at least) appears to be doing pretty well and a big part of that is investment in AI. We are building data centers, refurbishing fossil fuel plants to power them, and funding the tech entrepreneurs. This is a non-trivial stimulus to the U.S. economy and also fueling stock market growth, led by the tech industry giants. If the E.U. were to wholeheartedly embrace Vance’s offer it would mean expanded access for U.S. tech in the E.U. and expanded energy sales to the E.U. Of course, these are sectors that have heavily supported Trump and are benefitting from his policies. (Yours truly fears there is a lot of smoke and mirrors in the AI hype, but I claim no particular wisdom.) These constituencies were also an audience. (There’s probably a crypto play somewhere in there as well.)

Is non-ideological AI possible?

There was a tangled and contradictory thread in Vance’s remarks. Vance states, “AI must remain free from ideological bias, and that American AI will not be co-opted into a tool for authoritarian censorship.”

Not to get all post-modern, but non-ideological AI is not so easy in practice. There are certainly applications of AI that have limited ideological implications—using AI to design more efficient jet engines for example. But let’s take a simple example, using AI systems in a factory or warehouse. The machines can perform faster than people, so people might be injured trying to keep up. The decision to optimize to the machine rather than to the people is an ideological one—increased productivity is the ultimate good. Maybe the designers do a cost-benefit analysis and determine that paying for injuries is expensive, so they should change how the machines function. Alternately, they could do a normative analysis and decide that running a factory with machines that hurt people is morally wrong. All of these represent worldviews, systems of ideas, ideology.

Developing effective AI is complicated and involves lot of issues including the purpose of the AI, but also the data used to train the model. All of these things have assumptions baked in. Saying there is no ideology is silly. Instead it is better to examine the assumptions and be clear-eyed about how they shape the end product.

It was John Maynard Keynes who said, “Practical men who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct economist.”

This perfect, neutral AI is a shibboleth.

Does Vance really want non-ideological AI?

This problem becomes even knottier. Vance calls for limited AI regulation, but then VP Vance states:

In the U.S. we had AI image generators trying to tell us that George Washington was black, or that America’s doughboys in World War 1 were, in fact, women. Now we laugh at this now, and of course it was ridiculous, but we have to remember the lessons from that ridiculous moment that we take from it is that the Trump administration will ensure AI systems developed in America are free from ideological bias and never restrict citizens’ right to free speech. We can trust our people to think, to consume information, to develop their own ideas and to debate with one another in the open marketplace of ideas.

This paragraph seems a bit muddled. Vance is concerned about the fake images, but also says free speech is paramount. In that case, would one be free to make an “ideologically biased” AI? Could one develop an LLM which gives answers rooted in Marxist philosophy? What if one invents a DEI-AI and markets it as a decision support tool? Are these permitted endeavors?

Is Vance’s idea of non-ideological AI a Trojan horse for Vance’s preferred ideology?

Vance is a thoughtful person concerned with the clash of ideas. The cited paragraph seems mostly about social media, which is (as I wrote recently)³ central to the administration’s power but only one of many applications of AI. Is it possible that Vance doesn’t really care about AI as technology, but is more interested in AI and speech regulation.

We’ll look into that more when we discuss his Munich speech tomorrow. Stay tuned.

1. In fairness to Vance, he has allied with FTC chair Lina Khan in calling for antitrust measures against giant corporations. This is, however, a form of regulation.

2. American AI may be the best, but the administration Vance serves seems pretty hostile to basic research and high ed—the very things that put U.S. AI, and science and technology in general, in the lead.

3. In the post I wrote on how Trump and Musk have used their social media mob to cow Senators. But it’s also been turned on private citizens, who have less ability to withstand the onslaught. (It’s going to get people killed.)