Print
If you can’t beat ‘em, make a bot that can. That seems to be the mantra that two UC Santa Cruz students -- Ben Weber and Peter Mawhorter – are following in their attempt to design a StarCraft A.I. capable of going toe-to-toe with even the best of the pros.
“Our goal is to create compelling new forms of interactive art and entertainment that provide more deeply autonomous, generative and dynamic responses to interaction. A major thrust of this work is advanced A.I. for video games, including autonomous characters and interactive storytelling,” they told GamePro.
Currently, the crafty automaton’s already enjoying a 20 percent win rate against humans. It also plays exclusively as the Protoss, meaning that even if Weber and Mawhorter’s experiment fails, they’ve got an easy cop-out. “We were trying to make it play like a Pro all right. A Pro-toss!”
Er, anyway, the two hope their tiny tot of a bot will eventually reach “human-level behavior” – not just in individual games, but in the way it adapts to different situations.
"New strategies are constantly being uncovered by players and playing competitively requires learning how to counter newly discovered strategies," Mawhorter explained. "Our approach to this problem is to build a bot that learns new strategies from replays. Therefore, by giving it more replays, you can expand the strategic possibilities that it considers."
The plan, then, is for the bot to eventually learn strategies on its own, though implementing that kind of sponge-like soaking ability on anything more complex than build orders has proven to be quite the headache. We wish Weber and Mawhorter the best of luck. A word of advice, though: Make sure your bot stays far away from those illegal betting rings.
Comments are closed on this article
JohnP
April 16, 2010 at 12:12pm
On most RTS games, the computer beats me already on anything except easy. I want to win every once in awhile. Please!
