BBC NEWS Americas Africa Europe Middle East South Asia Asia Pacific Arabic Spanish Russian Chinese Welsh

 You are in:  Entertainment: New Media
Front Page 
UK Politics 
TV and Radio 
New Media 
Talking Point 
In Depth 

Commonwealth Games 2002

BBC Sport

BBC Weather

Saturday, 24 March, 2001, 11:50 GMT
Computer games get smarter
Creature from Black and White Electronic Arts
A lion goes bad in Black and White
By BBC News Online technology correspondent Mark Ward

What's black and white and played all over? No, it's not penguin polo, it's one of the most anticipated computer games of the last few years.

The game, which goes by the name of Black and White, is the brainchild of Peter Molyneux, one of the industry's most creative makers of games. He was the driving force behind such classics as Populous, Dungeon Keeper, Syndicate and Magic Carpet.

It is easy to design an AI-driven player that cheats, that can do anything you can but faster

Peter Molyneux, Lionhead Studios
Black and White has everything a modern computer game needs. It looks and sounds as good as it should, has an engaging story, is easy to learn but hard to master, and has hidden surprises that will keep you playing it for so long that significant others will give you up for dead. The game is due for release on 6 April.

But the most intriguing aspect of Black and White is the efforts its creators have made to make the creatures within it artificially intelligent. It is one of a growing number of games using techniques from artificial intelligence (AI) research to add depth to the software.

Intelligence included

Game makers are turning to AI to make titles stand out in the highly competitive world of computer games. They like it because, if it is done well, it is a very cheap way of adding playability. Instead of spending huge sums on sumptuous graphics, long cinematic sequences, endless levels, characters and plots, a game can be made much more challenging with a bit of nifty programming.

But putting AI in a computer game is not simple. Done badly, it can make it laughably easy to outwit computer-controlled opponents.

In the past, many game designers have used simple rules to govern the behaviour of the goblins, imps, aliens or soldiers that populate their software world. The rules tell these foot soldiers what to do if the human-controlled character or forces are nearby.

Badly thought out rules can make the aliens and soldiers react unrealistically. Rather than sounding an alarm and rushing to investigate as soon as one of their number disappears, a large group of combatants can often be tempted away one by one and slowly overcome.

Other games have used AI to make a computer-controlled opponent harder to beat by letting them cheat. With faster reactions, intimate knowledge of the game and troops that are better equipped, a game can be made formidably hard.

Quake II Id Software
Big guns and big brains make Quake tough to play
"It is easy to design an AI-driven player that cheats, that can do anything you can but faster," said Mr Molyneux. "The real trick is to make them do exactly what you do and instead of making them insanely hard to beat, making them entertaining and challenging."

Games such as Half-Life have been praised for their use of AI that makes human and alien opponents catch you out by working together rather than rushing in with all guns blazing. They also make inventive use of the weapons they carry, and have been known to use grenades to flush out a human-controlled character to drive them towards a trap or into withering cross-fire.

Using an AI engine called Soar US researchers have made a version of Quake that endows the computer-controlled opponents with more intelligence, making them tricker to overcome.

Newer titles like conspiracy game Majestic use AI to learn about a player's gaming style and tailor the story and interaction to suit. Majestic puts the player in the middle of an X-Files-type conspiracy and is driven along by supplying clues and interaction via phones, faxes, SMS messages and hidden documents scattered around the web.

Good God, bad God

But Mr Molyneux and AI whizz Richard Evans are trying to go further than this with Black and White. Playing the game involves taking on the role of a God and trying to gain dominance over a virgin world. Your Godly powers increase when you convince the inhabitants of the world's villages to worship you.

The game is riddled with AI. The villagers are smart enough to look after themselves by cutting wood and growing crops. They will even play a game of football if you provide them with a pitch. But they do best if you gently manage them and give them good reason to worship you by helping them or showing them the power of your wrath.

The world of Black and White is smart too and gradually changes to reflect your playing style. If you are an old-style vengeful God that demands worship and wreaks vengeance on those that don't unquestioningly accept your rule then the world will become darker, the landscape craggier and the sky more overcast.

If you are a benevolent deity who loveth rather than chastizeth then pastel shades will start to dominate and the flora and fauna will have more of a fairytale look.

Creature feature

But Mr Molyneux and Mr Evans have saved the most sophisticated AI for the creature that becomes your living representative in the world of Black and White. There are 14 creatures to choose from in the game, including apes, cows, tigers, wolves, lions, mandrills and zebras.

Zebra Electronic Arts
A zebra looks on in Black and White
Your creature grows and changes to reflect your Godly style and learns the spells that you master. But it is not a simple mimic and has been given a complex mental life of its own.

"The creature's actions are modified by what it has learned, by what it has seen and by its beliefs," said Mr Molyneux. "It's completely open-ended."

The creature has been given beliefs, desires and intentions that interact to let it learn relatively sophisticated concepts and turn learning from such things as "helping good, killing bad" to "planting trees for villages we want to convert is good, ripping up opponents' forests is good too".

This ability to discriminate has already produced some real surprises for testers and creators alike during development. In one situation, there was a creature that kept losing a stone-throwing game to another creature. To get revenge, the first creature heated a rock, planted it in the pile of stones to be thrown and then fell on the floor laughing when its rival burned its hands. All this happened without intervention from the player.

There doesn't sound anything artificial about intelligence like that.

See also:

20 Dec 00 | Health
Computer games pose injury risk
17 Oct 00 | Sci/Tech
Return of the computer dinosaurs
15 Nov 00 | Sci/Tech
Creating your cyber-self
03 Jan 01 | Sci/Tech
Hackers kill off heroes
28 Feb 01 | Sci/Tech
Computer 'can talk like a baby'
30 Aug 00 | Sci/Tech
Robot learns to reproduce
Internet links:

The BBC is not responsible for the content of external internet sites

Links to more New Media stories are at the foot of the page.

E-mail this story to a friend

Links to more New Media stories