In the competitive video game industry, successful game designers must offer savvy players ever more complexity. Now, devilishly smart enemies are taking on a life all their own.

The master chief of the U.S. Marines hoists his plasma rifle to his machine-hybrid shoulder. He takes a deep breath, then peers around the corridor of the alien space ship he’s infiltrated. His hand is on the trigger, ready to fire at the slightest flicker of movement. The way is clear.

Satisfied, he runs down the hallway. Suddenly a bolt of energy hits him, staining his vision red for an instant. He tries to spot the shooter, but more plasma fire hits him first. A light on his forearm console flashes yellow – his shield is reaching critical.

There must have been five, six aliens there, waiting to ambush him. Their attack was like nothing the human player controlling the master chief had ever experienced. The bad guys are getting smarter every year.

This frustration is what makes the computer game fun. It’s a concept called “gameplay,” the experience of playing a game. It includes perspective, the environment, how the game is controlled, and the enemies the player must vanquish. A satisfying gameplay is what drives a successful game.

With the video game industry now grossing as much or more on new releases as Hollywood blockbusters do on opening weekend, designers are eagerly pursuing ways to make games better. The key to this, many think, lies in artificial intelligence. Only A.I. can generate the rich and complex worlds gamers increasingly seek. They want intelligent characters that act autonomously and even emotionally. They want gameplay that adapts to their playing style and ability on the fly. All this is coming, and more, as a generation of young players, some even holding degrees in game design, become both the consumers and producers of the next generation of games.

Program developer Colin Moore of San Leandro, California, is typical. He says games that have realistic environments and smart opponents engage him and keep him playing. “It’s the unpredictability – it’s not like the movement is scripted. Sometimes they’ll attack, other times they’ll dodge if you try shooting at them, and in a lot of cases they’ll even run away.” He admits to playing three to four hours a day and recognizes his hobby borders on obsession. “Playing against the computer is almost as satisfying as playing with other people.”

Elements of the game

If the goal were simply to make the game hard to beat, the solution would be easy. Since computers are so much faster than people, a virtual enemy could be programmed to attack more quickly. The designer might also stack the deck in the enemy's favor with frustratingly 'unreal' powers like X-ray vision or uncanny hearing. Some games already do this, and the result is the bad guys are already shooting at you when you come around the corner.

These techniques might work some of the time, but the savvy player can usually tell when the computer is “cheating.” It feels, well, cheap. While such tricks may make the game hard to beat, they don’t necessarily make the gameplay better. The player gets frustrated with a computer that is simply faster or has unrealistic abilities.

“I hate when the A.I. cheats,” exclaims computer scientist Jim Whitehead. He is one of the creators of a new undergraduate major in video game design at the University of California, Santa Cruz. Instead, as Whitehead tells his class of over 100 students, “A good game offers choices, with no choice being obviously better than another.”

Tall and bespectacled, he gives his lectures with a smile pulling at his lips. It’s hard to imagine his lanky frame folding into a chair before a television set or computer monitor to play games, and yet that’s exactly what he does – for his research and for fun. He has two favorite games: Zelda, a Nintendo saga about an elf who must rescue a princess, and Civilization, a computer game of developing nations.

He likes these games precisely because they offer compelling choices. In Zelda, the gamer plays Link, a magical elf who must navigate the world of Hyrule. The graphics are rudimentary, with two-dimensional brown blocks for stones and green wavy blocks for trees. The original version is now 20 years old, but the game is a classic. The compelling elements are all there. The player isn’t forced through the levels but can play at his or her own pace, and choose what part of the world to explore first. The enemies, though simple, are worthy foes – skeletons with magical swords, ogres that throw spears, and fire-breathing dragons. They aren’t too easy to beat, but they aren’t frustratingly difficult either.

With more choices, the player has more freedom to explore, and the more involved the arena, the more satisfying the gameplay. As the gaming industry matures, the gameplay will advance, according to Whitehead. Far from the traditional fighting and combat scenarios, new video games offer players complex emotional and imaginative realms to explore. Some even allow players to create a world of their own.

A Virtual Microcosm

One way game developers introduce choice is through the game’s artificial intelligence. A.I. makes the game react to the player’s actions. If the player attacks an enemy, the A.I.’s program makes the enemies either counter attack or flee.

Halo 2, a creation of Bungie Studios in Redmond, Washington, broke new ground because of its well-developed A.I. The game was the much-anticipated sequel to the innovative Halo: Combat Evolved. The goal seems straightforward: You must save the earth from alien infiltration. It’s a common enough theme to be generic. And yet, within 24 hours of its introduction in fall 2004, Halo 2 sales exceeded $125 million worldwide. In less than a month, it had sold an estimated 5 million copies.

Why was Halo 2 so popular? Partly because of the enhanced graphics, but more importantly, the gameplay was real.

To create a virtual world, game developers must come up with a set of rules to define what can and cannot happen. Some rules describe the world and how things move within it, creating an environment that mimics the physics of real life. Other rules control how the player, enemies, and other characters within the world can interact with each other. These rules create the game’s microcosm. The more elaborate the rules, the more realistic the world. But at some point, it also becomes limiting – a game with too many rules doesn’t allow the player to act freely.

This is where the A.I. really comes into play.

A trademark of Halo and Halo 2 was in creating challenging enemies without relying on a “cheating” computer. Instead, Bungie Studios focused on developing the aliens’ A.I.s, according to Damian Isla, one of Halo 2’s A.I. designers. The A.I. controls more than seven different alien species, each endowed with varying amounts of aggression, agility, strategy, and targeting ability. Some scare easily, cowering before an aggressive player, while others are frighteningly malicious and surprisingly smart. It’s almost as if they can think for themselves – and that, of course, was Isla’s goal.

A Life of Their Own

Isla, 28 and a native of Peru, doesn’t play Halo 2 like a normal person would. “I’m probably the most atypical player,” he admits somewhat sheepishly. Unlike most players, he doesn’t engage the enemy. Rather, he likes to sit back and watch the game unfold. “I want the A.I. to do it all itself,” he says. “You’ll see a whole story being told that will never repeat itself. It’s really, really cool to see how they all come out.”

This isn't just idle curiosity. He’s watching for “emergence,” a general term for any way the game functions outside of its original design. Emergence can be initiated by a player interacting with the environment in unforeseen ways. For example, players of Halo 2 quickly realized that they could launch themselves into the air with an exploding grenade, a feature the developers had not foreseen. Similarly, a well-developed A.I. will play out unexpected scenarios. In both Halo games, the player is assisted by marines with an A.I. of their own. Mostly, they’ll attack any aliens on sight. As the game progresses, a player can sit back and watch the marines and the aliens fight as unscripted battles emerge from the rules of the game. These mini-battles can be as compelling as any action of the player.

The more robust the A.I., the more elaborate these scenarios become. But a complex A.I. program eats up a lot of the game hardware’s processing powers. The tradeoff for complexity is speed, and, for fast-paced games like Halo, loss of speed means dull play. If the aliens are sluggish, it doesn’t matter how smart they are – they’ll be sitting ducks.

Halo 2 avoids some of those issues by reducing the “cycle time” of the aliens’ response. Each time the player moves, he or she creates a new challenge for the aliens. Instead of tracking every step the player makes, the alien may wait until the second or third step. This makes for dumber aliens – they’re shooting in the place you were after you’ve split – but it cuts down on processing times. For the most part, it still happens fast enough to make the gameplay responsive.

Emergence as the new game

Most games, including the Halo series, have settings that can adjust to a player’s skill in an incremental way – easy, normal, hard, and “insane.” Usually games increase in difficulty only after the player completes a level. But many players would prefer a game that could continuously match wits with them as they improved. Now, there may be a solution.

It’s a program for the A.I. called “dynamic difficulty adjustment”, created by computer scientists such as Robin Hunicke of Northwestern University. Hunicke’s system changes the game’s difficulty to respond instantaneously to the player’s ability. While the player explores the game, the program assesses his or her ability to control the character and combat enemies, then alters the game to keep it challenging but not frustrating. Hunicke hopes similar systems will make action games more engaging and specific to a gamer’s abilities.

At 32 with blue eyes and dyed red hair, Hunicke represents a new age of video gamers, Ph.D. and all. She’ll talk incessantly about video games, given half a chance. Indeed, video gamers aren’t just teenage boys. More women are gaming now than ever before, representing almost half of all gamers. Their average age is 30, and most have played for the last 10 years. Video games have become a part of their lives.

This new generation wants to do more than fight aliens. “The best games are the ones where the player’s choices really drive the game forward,” Hunike says. For her, the ultimate game is an intricate world presenting the player with an infinite number of open-ended choices – essentially a game with a dynamic A.I. that would sculpt the entire game as the player proceeds.

Most action games like Halo can’t handle the complexity of such a dynamic A.I. But others, like Electronic Art’s The Sims, are designed solely to create an emergent world. Thousands of players can interact on The Sims through the Internet to create a customized virtual world. The rules of the game create as intricate a world as possible, defining the environment and the depth of interaction for the players. The point of the game is not a mission or an enemy to fight. Rather, it’s to live and interact with others in a virtual world that the players themselves shape.

Electronic Arts, based in Redwood City, California, is pushing the industry’s frontiers with its next creation, Spore, by designer Will Wright. The game is still under wraps, scheduled for release in late 2006. Already, the game has a reputation for being ambitious and innovative, a harbinger of a new generation. Gaming magazine IGN hailed Wright as “a mad genius of gaming” last May.

Spore depends on emergence, and its purpose is nothing less than to play with evolution. “I didn’t want to make players feel like Luke Skywalker or Frodo Baggins,” Wright said at a conference in 2005. “I wanted them to be like George Lucas or J.R.R. Tolkien.” The player decides how to evolve from a microbe into a space-exploring species. The program must create, from scratch, a response to the player’s choices. For example, if the player creates a creature with six tails, the computer has to put those together and decide what a six-tailed creature would look like, how the tails would join together, and how the creature would move. The potential for play and for exploring such an emergent world are virtually limitless.

Toying with emotions

If the current trend is freeing the interactions within virtual worlds, then the next unexplored territory is the emotional game. The experimental game Façade, created by computer scientists Michael Mateas of Georgetown University and Andrew Stern of Proceduralarts.com, is about human emotions and relationships. Like Spore, the mission is not to defeat a specific enemy but to explore an emergent world – with virtual humans. In this game, you are a good friend to a couple, Grace and Trip. Successful and in their thirties, the two are having a fight during an evening together, and it is your task to try to patch things up.

The scene begins outside their apartment, where the player can overhear a muffled argument through the door. Once inside, the player can navigate through the apartment, picking up objects, such as a drinking glass, and talk to the characters by entering text into a field. The game happens in real time – if you talk while Grace is talking, you interrupt her, maybe even angering her in the process. Nor will the characters wait for the player to make a move. Rather they will take off in a direction all on their own. Mateas and Stern’s goal was to create a seamless interactive narrative, incorporating emotion, dialogue, and action into a “one-act drama.”

To make the game work, Mateas and Stern developed four unique languages to interpret the dialogue and elicit and convey the appropriate responses from the characters through their facial expressions, their emotional reactions, and their verbal responses. Even if the player has entered something cryptic, the characters won’t say they don’t understand. Rather, the A.I. tries to detect words in the text to generate a comprehensive response. The characters will also respond verbally to an action. If the player makes him or herself a drink without a prompt, Trip may say he doesn’t feel like one right now. It can be frustrating at times – Grace and Trip get so engrossed in their argument that the player feels neglected. But most players are swept away in the relationship’s drama, playing again and again to determine how they might change the outcome. There is nothing else quite like Façade, but there probably will be more games like it. It’s “the future of video games,” according to a June 2005 article in The New York Times Arts section.

Until the future comes, the master chief can get away with simply shooting the aliens and trying to outsmart them. Soon, he might have to sit down and talk to them, even convince them to surrender. It’ll be a much tougher game, then, and someday, he might even start to feel bad for the guys.

ABOUT THE WRITER

Anne Pinckard
B.A. (integrative biology), University of California, Berkeley
Internship: Lawrence Berkeley Laboratory

As an undergraduate, I relished the opportunity to explore as many subjects as caught my interest. My studies encompassed such diverse subjects as medical ethno-botany, biochemistry, rangeland ecology, epidemiology, and genetics. As graduation neared, though, I was filled with a sense of panic. Too soon, I knew that I would have to answer The Big Question: what do I want to do with my life? My answer was always to wonder how could I possibly choose one subject to devote myself to for the next five years in a PhD program. Forget it, I decided. I would not choose. That’s when I came across the Science Communication program. Here, at last, was a career in which I could indulge in anything that struck my fancy.

ABOUT THE ILLUSTRATOR

Jennifer Marie Bolen
B.S. (molecular, cellular, and developmental biology), University of California, Santa Cruz

Since Jenna was little tromping around in her backyard or the Sierra Nevada she has enjoyed drawing what nature has to offer be it her cat or a redwood tree, yet she always had an equally strong passion for science. As a student she felt that she could earn her degree in MCD Biology and continue to draw as a hobby thus satisfying her love of both art and science. While at community college she worked as a teacher’s assistant and a Calculus Supplemental Instructor and in doing so learned she also has a flare for teaching. To her, the Scientific Illustration Program is a chimera of her three great passions and she hopes it will help her enter a field where she can be a hybrid of scientist, artist, and teacher.