How a Group of Marines Got Past AI-Powered Security



Artificial intelligence can do a lot. Given a large chunk of data, it can process information faster than your average intelligence analyst can. However, when it comes to some things, AI is apparently lagging behind Marine Corps intelligence. In just one day, a group of marines figured out how to sneak around and avoid detection by an AI-powered security system. They figured out that they best way to approach a system designed to identify human beings is to……..well…….not look like a human. As Phil Root, Deputy Director of the Defense Sciences Office at DARPA, pointed out, “A tank looks like a tank, even when it’s moving. A human when walking looks different than a human standing, and a human with a weapon looks different.” So, to train the artificial intelligence, a squad of Marines spent six days walking around in front of it. On the seventh day, it was time to put the machine to the test. Two marines somersaulted for 300 yards to approach the sensor, while another pair hid under a cardboard box. “You could hear them giggling the whole time,” said Root. Another Marine stripped a fir tree and held it in front of him as he approached the sensor. In the end, artificial intelligence couldn’t detect a single one of them. The lesson learned: AI can be very good at what it knows how to do, but AI doesn’t know what it doesn’t know, and it also doesn’t know that it should know what it doesn’t know. When it comes to Marines against AI, Marines remain undefeated.