András Lörincz and István Szita at Eötvös University in Budapest, Hungary have developed a computer program that outplays the average human at the classic arcade game, Ms. Pac-Man [1]. Starting off with scenarios (“ghost nearby”) and possible actions (“move away”), the program “learned” to perform better.
Should we feel inadequate now that a computer can beat us? Probably not. The program, as good as it is, could not learn some strategies that human players find very helpful. For example, most human players wait until a ghost is in range before gobbling up a power dot.
Still, in conjunction with Big Blue’s success in playing chess, we have to respect the remarkable and rapid improvements in artificial intelligence. I’m still waiting for a robot that will clean my house, though, while I play Ms. Pac-Man.
1. Szita, I., & Lorincz, A. (2007). Learning to play using low-complexity rule-based policies: Illustrations through Ms. Pac-Man. Journal of Artificial Intelligence Research, 30, 659-684.
3 Comments
Vix · January 22, 2008 at 10:40 pm
Since I’m the lone gamer of the family most of the time, I’m always undefeated when it comes to video games, as no one beats me 🙂
ccallag · March 13, 2008 at 2:02 pm
I love pac man! Thats one of the few video games I will play. I think its eerie that the computer beats the human every time. I always think of the movie iRobot when it comes to artificial intelligence. I don’t know if its such a good idea making computers smarter then humans. yikes!
Amber Thompson · October 19, 2008 at 7:32 pm
If you read the article correlating to this blog, ‘ccallag’, you’d see that the computer doesn’t beat the human every time, but beats the average human. What I wonder is how much the computer can “learn”. Can it pick up these more advanced strategies, if “taught” to recognize them and apply them?
Our brains are set up much like computers, with signals that are sent to different parts of the body and “memory storage” for “filing” different information in different parts of the brain. I have no doubts that in time, we will be able to create computers that have the ability to “learn by experience”.
Comments are closed.