Game Development Reference
In-Depth Information
chapter 1
What Is Game AI?
Our working definition for AI is the ability to act intelligently in the face of
changing conditions. Embedding such capability in a game makes it game AI. We
will return to this definition throughout the topic to make sure that we can
identify how any given AI meets the definition. We will also take care to examine
what is not game AI. While this definition seems simple enough, all three parts of
it—the ability to act, the requirement that the action be intelligent, and the
requirement that the action be in response to changing conditions—are worth
separate inspection.
First, as mentioned, the AI should have the ability to act in some fashion. AI that
has no output is wasted computation. In a game context, this means that the
actions of the AI must have the possibility of being noticed by the player. Games
that are noted for good AI often have an AI that is no better than other games,
aside from the fact that the game gives the player the ability to see the AI acting
intelligently. For example, if, in a game, the player sees a room with an AI-
controlled character working at a computer, the player may conclude there is
nothing special about the AI. Even if the AI character happens to be playing a
world-class game of chess in real time, it is all wasted unless the player gets a
chance to notice this amazing ability of the AI. The player does not necessarily
have to see the intelligent behavior at the time a decision is being made, but
evidence of the intelligence should be made available to the player in time for the
player to enjoy it.
Search Nedrilad ::

Custom Search