The Pentagon’s Defense Advanced Research Projects Agency (DARPA) has invested some of its resources into a robot trained to identify humans. There’s just one little problem: The robot is cartoonishly easy to confuse.
Army veteran, former Pentagon policy analyst, and author Paul Scharre is gearing up to release a new book called Four Battlegrounds: Power in the Age of Artificial Intelligence. Even though the book isn’t scheduled to hit shelves until Feb. 28, Twitter users are already sharing excerpts via social media. This includes The Economist‘s defense editor, Shashank Joshi, who shared a particularly laughable passage on Twitter.
In the excerpt, Scharre describes a week during which DARPA calibrated its robot’s human recognition algorithm alongside a group of US Marines. The Marines and a team of DARPA engineers spent six days walking around the robot, training it to identify a moving human form. On the seventh day, the engineers placed the robot at the centre of a traffic circle for the Marines to approach from a distance and touch the robot without being detected.
DARPA was then humbled. Scharre writes that all eight Marines could defeat the robot using techniques that could have come straight out of a Looney Tunes episode. Two of the Marines somersaulted toward the centre of the traffic circle, a form of movement the robot hadn’t been trained to identify. Another pair shuffled toward the robot under a cardboard box. One Marine even stripped a nearby fir tree and reached the robot by walking like a fir tree (how do fir trees walk?).
While it’s funny to imagine a team of Marines using a cardboard box strategy to defeat a very expensive robot, the incident confirms what we shouldn’t forget; AI is only as useful as the data we give it. Similar to how AI becomes biased once it’s fed biased data, algorithms can be as ignorant as their foundational and data is flat. Without being shown what a somersaulting human or a human under a box looks like, a robot won’t be able to discern that image from all the surrounding noise, no matter how skilled its engineers are.