[Image: Courtesy Iris Automation].
In an interview published on the blog here a few years ago, novelist Zachary Mason, author of The Lost Books Of The Odyssey, pointed out something very interesting about the nearly limitless, three-dimensional space of the Earth’s atmosphere and how it relates to artificial intelligence.
“One of the problems with A.I.,” Mason explained back in 2010, “is that interacting with the world is really tough. Both sensing the world and manipulating it via robotics are very hard problems, and [these are] solved only for highly stripped-down special cases. Unmanned aerial vehicles, for instance, work well because maneuvering in a big, empty, three-dimensional void is easy—your GPS tells you exactly where you are, and there’s nothing to bump into except the odd migratory bird. Walking across a desert, though—or, heaven help us, negotiating one’s way through a room full of furniture in changing lighting conditions—is vastly more difficult.”
Another way of thinking about Mason’s comment—although Mason himself might disagree with the following statement—is that it is precisely the sky’s ease of navigation that makes it ideal for the emergence and testing of artificial intelligence. The Earth’s atmosphere, in other words—specifically because it is an unchallenging three-dimensional environment—is the perfect space for machine-vision algorithms and other forms of computational proto-intelligence to work out their most basic bugs.
Once they master the sky, then, autonomous machines can move on to more complicated environments, such as roads, mountains, forests. Cities.
In any case, I was thinking about Mason’s interview again earlier today when I read that drones are close to achieving “situational awareness”—albeit through visual, not artificially intelligent, means. In other words, it’s not AI—at least not yet—that will give unmanned aerial vehicles their much-needed ability to avoid colliding with other flying objects. Rather, it is a sufficiently advanced visual processing system that can identify and, more importantly, avoid potential obstacles.
Exactly such a system, TechCrunch claims, has been built by a Canadian firm called Iris Automation. Their system is able “to process visual data in real time, so it can see structures that suddenly appear, like a plane, flock of birds or another drone—not just static objects and waypoints that might be mapped using older technologies like GPS.” The company refers to this as “industrial drone collision avoidance,” which suggests a kind of on-board traffic management system for the sky. Air traffic control will be internal.
Now connect a drone’s “situational awareness” to sufficient processing power, and you could help steward into existence a computationally interesting form of autonomous intelligence.
To return to Zachary Mason’s computationally-inflected rewriting of The Odyssey, it would be AI as Athena, springing fully formed into the world from an empty sky.