My thinking is that if we want to avoid such a fate, the first place we need to start keeping a close eye on is in our games.
Ever since the early single player games -- games like Pong, Space War, and the like -- we've fought our battles against the very machines that held the games we enjoyed. Though they may have taken on the identities of aliens, monsters, and evil wizards, our real opponents all along were the computers embedded in the console. So we clearly have a long and storied history of taking arms against artificial intelligence, even in this rudimentary form.
Consider the ever-growing popularity of MMOs. Even though PvP does account for a good chunk of online play, there is still a sizable contingent of players who enjoy the challenge of leveling up, and taking on the neverending mobs inherent in these worlds. And all of these enemies are just instances of artificial intelligence, intent on destroying the player. Even worse: networked artificial intelligence. Isn't this how Skynet got started?
Of course, our games are nowhere near the computational power of that fabled machine, but these things always begin small. Recall how WOPR nearly caused the devastation of Earth in Wargames -- a simple mainframe running tactical algorithms nearly took the world down accidentally. The very last thing we need is to add the element of free will to that situation.
And yet, isn't that the ultimate goal of game designers? To create an opponent that's smart enough to outwit the player? There is a long list of games that could benefit from smarter enemies. Think about the numerous ways to exploit the simple follow behavior of the robots in Loderunner, or the way that the soldiers in the Metal Gear Solid series manage to forget all about you if you hide away long enough. Sure, there were probably sound gameplay decisions made that resulted in these behaviors, but as the game playing audience matures, it demands that the genre mature along with it, and these days that means enemies that behave more like other players do in PvP.
To wit: enemies should intelligently take cover; learn to flank the player; make use of more than just one weapon; find and make use of health packs; generally do all the things that good players do themselves. But that does raise the question: would smarter AI make our games more fun?
That would seem primarily to depend on the genre of game we're playing. World of Warcraft, for example, works pretty well as-is; if indeed the whole point of WoW is to indulge in the thrill of leveling, then we don't really need or want the creatures we fight to be any smarter than they are. And even in the multiplayer boss battles, we rely on our ability to discern a pattern and figure out how to defeat it. In a real-time strategy game, however, where your tactics are the key to victory, nothing beats the satisfaction of being able to out-think the opposite side. Should you overwhelm the enemy with sheer numbers? Should you lure them into a boxed-in area with the promise of an easy victory? Should you concentrate on climbing the tech tree in the hope that more powerful units will turn the tide? And most important of all: which of these strategies is the AI employing?





