Try to battle the artificial intelligence that constantly learns from its mistakes

The new browser game Aim and Shoot gives a great idea of ​​how neural networks work, how they learn, and what they can do when given free rein. The programmer Victor Ribeiro developed it for this - any user can personally experience how artificial intelligence learns, how quickly it adapts to new situations and begins to surpass humans.

The gameplay is extremely simple and ascetic: units with cannons move across the empty playing space, from which they must shoot each other. There is a reserve of health, there is a pause for reloading the weapon, so as not to fire in all directions for nothing. It is quite possible to dodge enemy shells, plus the player is not the main target. This is a typical desmatch where the only survivor wins.

Initially, the neural network does not understand what to do at all, but as soon as the first blood is shed, the learning process rushes forward with leaps and bounds. At the end of each round, the two best virtual fighters, with the combat skills they have developed, merge into a new type of enemy. And the new round starts with the same number of gladiators, but already "pumped". The player can use any tactics - the AI ​​will constantly learn from its mistakes.