Abstrakt
The article addresses the gap in scholarship on reception of game AI by qualitatively analysing forum discussions about the Alien monster in the game Alien: Isolation (Creative Assembly, 2014a). The game's Alien is a rare example of a recurrent, undefeatable AI-driven adversary, designed to represent a sublime monster.
Its behaviour is governed by behaviour trees and simulated sensory perception (on the level of foreground AI), but also by the Director component, which monitors the global state of the game to maintain gameplay tension and periodically sends the creature near the player's location (on the level of background AI). Upon the game's release, however, very little information about its AI was disclosed.
The article follows players' efforts to theorize and evaluate the game's AI, and their discussion about its (un)fairness. It outlines two basic approaches to enemy AI: experientialist, which focuses on how the opponent "feels," and simulationist, which sees the monster as a simulated animal-like creature.
While experientialists accept AI tricks and shortcuts, simulationists expect the Alien to be a discrete, autonomous entity, ontologically separated from its environment. The latter, for example, find it unfair that the Alien has access to information outside of its sensory perception, and seem upset by the actions of the AI Director module, which they interpret as collusion between the Alien and the game.
Moreover, simulationists object to the idea that the Alien should be able to teleport or be "tethered" to the player character. The article points to an important paradox of AI in games.
On the one hand, players welcome a degree of unpredictability, variation and mystery afforded by AI routines. On the other hand, many expect to be able to completely account for and understand the behaviour of computer-controlled opponents -- even if they are science fiction monsters.