The U.S. Military Bets Big on Artificial Intelligence

The Pentagon is currently funding over 800 separate AI projects

A great deal is being written about artificial intelligence: how it’s ruining education, how it’s taking over entertainment, and how it’s going to forge an unforgiving surveillance state that will sniff out our every blip to get at what we’re up to and what we’re thinking.

The Pentagon, which is currently funding a whopping 868 separate artificial intelligence projects, as I reported in a new article for The Intercept, is gaga for AI. As the federal agency with the most money, it plays the most influential role in determining what the U.S. government is going to focus on.

When any new technology — somehow always a silver bullet that is going to vanquish America’s foes — emerges, it is ripe for waste and dead-end inquiries. The case of AI is no different. Much of the Pentagon’s interest in AI is to help it to solve a problem of its own making: the intelligence apparatus collects too much information, way too much for analysts and even machines to make sense of. AI is needed, the Pentagon now says, to allow those machines to “learn” what is in the vast quantities of data. This is in its infancy now, but that’s why it’s important to intercept the train before it picks up so much momentum there’s no stopping it, or more importantly, direct it to the right station.

But first the Pentagon has to figure out its own policies. As I wrote in The Intercept, tucked away in the Defense Department’s public statements and budgets is one AI program – already up an running, that sorts out what American national security policy actually is.

By the Pentagon’s own admission, the “byzantine” rules it’s supposed to adhere to have grown to such a size — including 15,000 separate policy documents(!) — that merely sorting through them has become a “daunting task.” I’m quoting from the Defense Intelligence Agency’s triumphant announcement of its AI program designed to sift through the military’s “mountain of policies and requirements” (again, their words). 

Naturally, the AI program is titled “GAMECHANGER.” Self-licking ice cream cone, meet self-patting back.

We are supposed to believe that sagacious warrior monks are at the top of the military, masters of all things military history and even combat veterans, in command (literally) of what the military needs, how it will use it on the battlefield, how America will win. Putting aside for a moment how empty the win column is, the military’s own convoluted mess of policies explode that myth. Everything from old laws such as the Privacy Act of 1974 or the Freedom of Information Act of 1976, to the latest National Strategies on this and that, to appropriations and White House directives and internal directives, regulations, instructions, memoranda, orders, and on and on require that things be done in a certain way.

That includes thousands of classified documents that regulate everything from how the intelligence agencies can spy to how many civilians it can kill on the battlefield.

AI is supposed to clean up this mess. The same generals who’ve lost basically every war since 9/11 are at such a loss to understand the military’s own policies that they had to develop a special artificial intelligence program to make sense of it all for them. Anyone want to take bets on how long it will take before they develop another AI program to make sense of the current one?

Next up: Chairmain of the Joint Chiefs of Staff Gen. Mark Milley’s retirement and what’s to celebrate.