You might have encountered aircraft controlled by artificial intelligence in video games, but they were nothing compared to a new AI developed at the University of Cincinnati. Not only can the AI known as Alpha take out all the other simulations thrown at it, even an experienced human pilot was repeatedly outmaneuvered. So, I guess this is how Skynet starts.To test Alpha, researchers set up a simulated combat scenario with four enemy aircraft attacking two Alpha-controlled jets. Alpha had no trouble taking out the enemy aircraft, so a more challenging opponent was needed. Retired U.S. Air Force Colonel Gene “Geno” Lee hopped in a simulator to go up against Alpha, but was unable to best the AI. In fact, his thousands of missions and decades of experience were of little help as he was repeatedly shot down by the AI. Lee calls Alpha, “the most aggressive, responsive, dynamic and credible AI I’ve seen to date.”The Air Force already uses some artificial intelligence systems in drone aircraft, but there are no 1-on-1 combat drones that are controlled entirely by AI. A jet fighter produces a lot of flight data that can be hard for an AI to interpret and incorporate into its decisions. Alpha manages to overcome the complexity of the task with a decision making system called a Genetic Fuzzy Tree (GFT), which is a type of fuzzy logic algorithm.Using the GFT makes Alpha extremely fast and doesn’t eat up a lot of computing resources. It considers its actions more like a human would by breaking complex tasks into smaller ones like when to fire, how hard to bank to avoid return fire, and what angle of approach to use. By looking only at the most relevant variables, it can calculate the best maneuvers for a given scenario much faster than a human opponent.The creator of Alpha says this process is so efficient it could run on a Raspberry Pi, but right now it’s running on an entry-level $500 consumer PC. Alpha might not be flying combat missions any time soon, but the Air Force is interested in using the AI for research purposes and training.