Nuclear Weapons and AI

Today, the Doomsday Machine in the movie “Dr. Strangelove: or How I Learned To Stop Worrying And Love The Bomb” would be called an artificial intelligence device. The movie remains relevant, as it has for over 60 years.

Today we have a Wired article that breathlessly tells us “Nuclear Experts Say Mixing AI and Nuclear Weapons Is Inevitable.” This is a true statement, for certain values of “mixing” and “inevitable.” It’s inevitable because AI proponents are forcing “AI” into everything. “Mixing” could mean anything.

So it’s possible, as it’s been for eighty years, that someone would build a more refined Doomsday Machine than that imagined sixty years ago. And that’s the hook that the article depends on, along with the conference with Nobel laureates. When I read sentences like these,

In the middle of July, Nobel laureates gathered at the University of Chicago to listen to nuclear war experts talk about the end of the world. In closed sessions over two days, scientists, former government officials, and retired military personnel enlightened the laureates about the most devastating weapons ever created.

I despair.

Nobel laureates have been recognized for particular contributions. They are generally smart people and probably are aware that nuclear weapons are bad. It might help to have them spread that message. The nuclear war experts quoted in the article are pretty much the same nuclear experts as usual. They have spent their lives thinking about such things.

There is currently a concern among nuclear experts that the public has lost sight of the dangers of nuclear weapons. So they are bringing out the means they have used for the last sixty years or so to enlighten the public. Maybe the public will listen to Nobel laureates, if those laureates can be convinced to take up the cause. We saw a months-long series of articles in the New York Times about how nuclear weapons are bad. I see nothing to indicate it has changed public opinion. And, in any case, what can members of the public do about nuclear weapons?

I think that some folks have pictures in their minds of the organizing against fallout in the 1950s, centered on radioactive isotopes in children’s teeth, and against nuclear weapons in the 1980s reappearing as demonstrations in the streets to end nuclear weapons. But I don’t see how that is likely to happen.

Public marches and demonstrations are not as popular as they once were. Arguing about why is outside the scope of this post.And people have other concerns, like the rewriting of the Constitution by the Trump administration. I know that I’ve prioritized the fight to save democracy over nuclear disarmament. Which doesn’t mean I’ve forgotten it.

The nuclear experts quoted in the article are part of an establishment that is unable to consider new approaches. They dismiss the Treaty on the Prohibition of Nuclear Weapons, rather than engaging with it. The problems they raise with the treaty are real, but if they are serious about ending the nuclear threat, a deeper engagement might be in order.

The concept of the conference is rooted in past failures to engage the public. Let’s add AI, which everyone has heard about!

I regret to inform them that we do not have AI. We do have Large Language Models, which the grifters of Silicon Valley want to sell as AI. They aren’t; they’re more like large weighting and averaging machines. As such, they are useful in very well-defined and carefully limited applications like interpreting medical x-rays.

Nobody knows how to get from those LLMs to AI except by using words describing human phenomena to describe what they do and by setting the machines up to confirm the users’ preconceptions. If it sounds like AI, it must be AI, right?

Silicon Valley is forcing their averaging machines into all their products, and some potential users, like university administrators and nuclear experts, don’t want to be left behind, so they talk about what that fictional AI-in-the-sky can do some undefined day.

The Trump administration has been happy to accommodate their Silicon Valley friends and wants to insert the averaging machines across the government. So far, their attempts to apply it via DOGE, for example, have been a miserable failure.

But yeah, let’s talk about AI for nuclear war.