AlphaZero's Artificial Intelligence software beat the best chess, go, and shogi (Japanese chess) software after only a few hours of learning, and by training against itself. This was in December 2017.
In the same vein, MBDA researchers believe that machine learning algorithms or other techniques of Artificial Intelligence can be exploited to improve the performance of defense systems.
Take the case of air defense. By describing to a machine the behavior of the missiles (attacking as well as defending missiles), it can unroll a very large number of simulations corresponding to scenarios in which the number of attacking missiles and defenders can vary in quantities representative of real confrontations.
Like AlphaZero, this machine can quickly become an expert in defense tactics, offer the best time to fire an interceptor missile or salvo, against a particular attacker, etc.
The goal is not to replace a human operator, who will always remain the ultimate decision-maker in a military confrontation. But, as with AlphaZero for the game of chess, we hope to discover effective tactics of defense against tactics of unforeseen attacks, and thus be able to provide decisive assistance to a human operator when he is confronted with complex attack situations.
This is the meaning of the Open Innovation challenge that MBDA has proposed, and which has been retained by BPI France along with eight others proposed by companies and key corporate accounts in the framework of the "Artificial Intelligence" Plan announced by the President of the Republic in March 2018.