Which computer game has the best AI

Why Artificial Intelligence (AI) is stagnating in games

1996: A harvester, extremely pixelated by today's standards, searches for the raw material Tiberium in the strategy leader Command & Conquer, but simply rolls into the middle of the enemy's headquarters and explodes in an ugly mini-explosion.

2007: A harvester at the highest graphic level searches for the raw material Tiberium in the strategy hit Command & Conquer 3, but simply rolls into the middle of the enemy's headquarters and is impressively torn into its individual parts by glaring laser beams, which fly physically correct in all directions . There are eleven years between the scenes described above. Eleven years in which optics have developed enormously: 3D graphics, special effects, physics simulation. Eleven years in which, however, one important detail has not changed: the harvester behaved just as unwise in 2007 as in 1996.

The game Command & Conquer is just one example of a fundamental phenomenon: The graphics of computer games are constantly being improved, but the simulation of artificial intelligence (AI) has been standing still at a low level for years. What are the causes of this widening gap between realistic worlds and realistic behavior?

This is how AI works

The Command & Conquer 3 harvester doesn't do anything wrong. It independently finds its way through complex areas to distant mining areas or automatically registers when a raw material deposit has been harvested. That is why the technical director at Related Designs and chief programmer of the strategy game Anno 1701 Thomas Stein sums up the thankless task of the AI ​​developers: "Artificial intelligence is usually only noticeable when it is not working properly." But how does an AI even work?

Simply put, the programmers define for all units, buildings, non-player characters, enemies, and so on, how to behave in certain situations in a game. The more such behaviors are defined in connection with certain situations, the smarter the artificial intelligence of a computer game appears.

A simple example: In the programming tutorial AntMe! from Microsoft (see picture), ants collect fruit under time pressure and fight powerfully superior beetles. An AI program code defines the behavior of an ant when it encounters a dangerous beetle:

public override void looks (beetle beetle)
{
if (number in sight distance> 10)
{
AttackAn (beetle);
} else {
Turn around();
GoStraight (100)
}
}

The following happens: As soon as a beetle comes into our ant's area, it checks whether there are more than ten other ants in its immediate vicinity. If so, it attacks the beetle. If not, she runs 100 paces in the opposite direction.