It's been ten years since the release of Black & White, Lionhead's critically acclaimed God-simulator designed to bring us one step closer to realistic artificial intelligence.
By concentrating on procedural choices made by the player that were then fed back into the environment, Black & White aimed to present a world in which gamers could explore the overarching effects of their actions. Likewise, developers could push the boundaries of artificial intelligence by examining the interaction between the player and in-game characters, and use the results to build personalities.
On a rudimentary level, basic player input was mapped using a reinforcement model. Should the in-game pet – referred to as the Creature – displease the player, swift justice would be enforced with a slap. Affirmation of tasks completed satisfactorily would be achieved with a stroke. It's a time-honoured way to teach the fundamental construct of any social group, after all.
Lionhead then went one further. By adapting a Belief-Desire-Intent structure, the developers were able to turn the tables on the development of the Creature by affording it a functional decision making process of its own. Rather than acting in a pre-set way governed by existing programming, the Creature could choose from a series of known values formed by player reinforcement, and actively rate the probable outcome that each one would allow it to fulfil a desire. Thus, a goal could be formed.
This approach is hardly revolutionary, nor particularly complicated. It could also be argued that it failed to elevate Black & White much past a ham-fisted attempt to capitalise on lead designer Peter Molyneux's glory days, where he reigned over such creative endeavours as Populous and Syndicate. Regardless, the thought that these concepts were being explored a decade ago raises the pertinent issue of modern game design, and why we're still waiting for characters that can amaze us with obvious signs of intelligence.
It's not that the expectation is unreasonable. It seems that unrealistic enemy activity is practically a design choice in many of the latest titles. In the minds of developers, any bullet impact that doesn't kill an enemy combatant instantly merely encourages them to fight harder, as opposed to writhing on the ground in agony and expressing loudly the desire to avoid further hostility.
Nor has it ever been remotely realistic to witness innumerable targets briefly breaking cover from exactly the same location to take pot-shots at your squad. We certainly appreciate the metronomic qualities imbued to the cyclical appearance of a large round head above a concrete barrier, and the resulting melon-slammed-on-the-pavement headshot routine is satisfying, but is this really the most realistic scenario available in game development?
If it's not a complete lack of intelligence that so frequently hamstrings the experience, it's the ridiculously skilled robot-like enemy that, at a distance of half a kilometre, can immediately detect the exact source of an incoming round, and return fire before the player can even duck to cover.
Shooter titles are hardly alone in this regard. Strategy games often feature opponent AI with inherent bonuses that are simply unavailable to the player. Rather than trying to program enemy AI to think, behave and play like a human, developers simply imbue them with increased hit points, better statistics, or any number of favourable benefits whilst removing the same attributes from the players themselves. Manipulating these variables commonly forms the structure of a difficulty level.
In the quest to present the player with an action-packed experience, developers often sacrifice realism to keep the story flowing, or keep the participant at the crest of a wave. It's unlikely, after all, that precise emulation of a Special Forces raid in war-torn Afghanistan would lend itself well to the gaming public. An oft-quoted observation has us believe war is 90% boredom and 10% sheer terror, and it's the latter that sells video games.
The recent release of Crysis 2 further reinforces the abject lack of attention paid to the development of artificial intelligence in modern games. This is a title that should be at the forefront of first-person shooter innovation, graphically and technically, and yet broken pathing routines, unrealistic shooting mechanics, and the lack of any appreciable flanking system places it on a par with titles from the last millennium. At no point during the play-through of this ultra-modern shooter – a title specifically marketed on its technological accomplishments – was the enemy able to present behaviour that might compel the player to assume they were more than automatons executing poorly conceived lines of code.
So much for the meta-world concept so doggedly pursued by Crytek, and others guilty of spreading their creative butter too thin on the proverbial toast. Their generalised focus, no doubt exacerbated by lofty ambitions of open-world freedom, counts for little when the artificial intelligence introduced is so woefully unrealistic as to become an anachronism in itself.
It seems that the last decade of game development sought to build the foundation for mainstream gaming, first with increasing graphical power, then the explosion of multiplayer gaming, and lately with novel new control schemes. Throughout that process, we witnessed a remarkable expansion in player numbers, and the general reach of gaming as an entertainment medium. This decade, it's time to refine and enhance the core attributes, the very structure of what makes for a compelling interaction with the product you've invested in. There's no better place to start than the integration of truly intelligent computer-generated foes. It's long overdue.