DARPA has approved the second phase of artificial intelligence technology to help automate air traffic control. Its initial purpose is to help the Air Force, in particular, keep airspace operating safely with increased air traffic with the advent of unmanned aerial vehicles (UAVs) and other airborne weapons. DARPA says the software will learn by assembling knowledge from different sources—including generating knowledge by reasoning.
The Generalized Integrated Learning Architecture (GILA) system, developed by Lockheed Martin, sounds like the beginnings of Skynet,
albeit 10 years later than expected. GILA will help create airspace control orders by automatically learning flight controllers’ tasks—often only using one example. According to a Military & Aerospace item, such software has to combine limited observations with subject expertise, general knowledge, reasoning, and by asking what-if questions.
What-if questions? Air traffic control is one of the most stressful jobs on the planet, and I’m all for giving controllers better tools with which to do their jobs. But what if a split-second life-or-death decision needs to be made which requires more information than the software has been given? What if the software is given bad data? What if air traffic control decisions require human reasoning, not just the black-and-white “reasoning” of a computer? Those are my what-if questions. Air traffic control is not something that should be left, ultimately, to artificial intelligence.
