ALPHA, an artificial intelligence developed by a doctoral graduate of the University of Cincinnati has beaten retired United States Air Force Colonel Gene Lee who tried his skills against ALPHA’s superior algorithms.

ALPHA also defeated all other AI’s available at the Air Force Research Lab and any other volunteer human experts. The contests were in a high fidelity air combat simulator.

The most notable aspect was that ALPHA achieved all of its superiority running on a Raspberry Pi.

The feat of defeating Lee was no mean achievement. Gene Lee, an instructor, is experienced in aerial combat having trained with numerous U.S. Air Force pilots. Other achievements by Lee are being an Air Battle Manager who has fought numerous AI opponents in air combat simulations since the 1980’s.

Not only was Lee unsuccessful in winning against ALPHA, but he also could not win even when ALPHA’s aircraft was impeded in speed, sensor use, missile capability and turning.

According to Lee, he was surprised at how ALPHA was reactive and aware. It knew how to defeat every shot as if it was aware of Lee’s intentions. ALPHA also reacted to Lee’s changes in flight and missile deployment. ALPHA seamlessly changed from defensive to offensive actions as required.

What sets ALPHA apart is the genetic fuzzy tree decision system that can calculate an opponent’s movements or strategies 250 times faster than you can blink. Such speed gives it a humongous advantage in a field where advanced skills and great intuition are required.

The future of Air Combat could be dominated by ALPHA which could be a valuable asset to team with a fleet of human pilots. It could help to map right strategies and coordinate a team of aircraft.

Kelly Cohen, an aerospace professor at UC, says that ALPHA could be used to determine the best possible way to perform tasks commanded by its wingman. It will also provide situational and tactical advice during flight.

However, ALPHA has not been without concerns such as pioneering the era of autonomy in battle aircraft. Eventually, we could see Unmanned Combat Aerial Vehicles (UCAVs) deployed in missions without human input and helping to eliminate human errors.

Nick Ernest, a founder of Psibernetix that developed ALPHA, says that they need to push and extend the capabilities of ALPHA. There need to be realistic aerodynamic and sensor models to increase fidelity. The extension of capability will be followed by additional testing of ALPHA against other trained pilots in the simulated environment. Psibernetix is ready to engage in continuing development.

5 COMMENTS

  1. “The future of Air Combat could be dominated by ALPHA which could be a valuable asset to team with a fleet of human pilots.”

    Wait, what? If you have a superior AI pilot available, why would you still have any human pilots at all?

  2. The US knows Russia has been building better new aircraft than they have and perhaps they hope AI flying US planes can beat Russian pilots in Russian planes. However, I would have thought even though AI can think faster than a Russian pilot it will still be limited by the limits of its own aircraft. For all its brains it cannot make an aircraft do what the aircraft cannot do. Those problems are down to aircraft ability not pilot ability.

LEAVE A REPLY

Please enter your comment!
Please enter your name here