
Getting this illusion of artificial intelligence right can make or break a game. “Solid AI provides excellent interactivity, gameplay and immersion into the world,” Kripalani opines.” AI is critical because characters have to be believable. You meet characters in these games. How they react to you influences your suspension of disbelief. You want to be convinced.”
And nothing could be more convincingly real as behavior that truly reflects real life, “A while ago we built dynamic avoidance into our system,” Kripalani tells us. “This is the ability of characters to walk in an environment without colliding into one another. It is a surprisingly complex problem. After we had the system up and running we noticed an emergent behavior. The AI characters in the system determined that it was easiest to walk in lines on either side of a tight hallway rather than randomly trying to cut in front of each other anywhere. If you think about it, this is how humans react when facing the same problem.”
In “the real world” a successful implementation of Artificial Intelligence could mean a safer way of doing things. “AI systems can perform tasks that may be too dangerous for humans, or even animals. Would you rather send a dog or a robot into a building to search for explosives?” asks Dr. Courte. More often, though, it could mean inferring solutions far too complex for humans to calculate.
“Other areas [of AI], such as data mining and evolutionary computation, take novel approaches to exploring large search spaces where an answer to some unanswered question may lie. The goal in this case is often information discovery, going beyond what humans have learned.”
Dr. Courte provides us with one example of data mining, “At convenience stores, a purchase of disposable diapers is frequently accompanied with a purchase of beer. I doubt that most store managers would have on their own determined that placing beer displays near the diapers would be a smart marketing move, but the data, properly analyzed, indicates otherwise.”
Discoveries made by artificial intelligence systems today are more and more complex as the technology develops. How long, then, will it be until we create an AI that is just as, or even smarter, than you and I? “Centuries and centuries,” thinks Kripalani. “This is an amazingly complex problem that involves issues of cognitive models, perception of emotion, language and a myriad of other issues... Do you think we have the ability to teach a computer what it means if someone does not complete a sentence? Everything's so situational it is mind-boggling to try to capture all of the nuances. I think we are a very long way off.”
As extreme as Kripalani’s stance was, Dr. Courte seems to agree. “I’m sure you can find articles in the popular press that predict that soon we’ll communicate with computers as if they were human, much as in Star Trek or other fictional worlds. But you can go back at least three decades and find similar predictions. So far, ‘soon’ has not come, and merely conversing with a computer that seems human is a long way from true intelligence.” He concludes, “I think there is a lot more to learn about human intellect and the brain before we can even begin to contemplate such systems."
And nothing could be more convincingly real as behavior that truly reflects real life, “A while ago we built dynamic avoidance into our system,” Kripalani tells us. “This is the ability of characters to walk in an environment without colliding into one another. It is a surprisingly complex problem. After we had the system up and running we noticed an emergent behavior. The AI characters in the system determined that it was easiest to walk in lines on either side of a tight hallway rather than randomly trying to cut in front of each other anywhere. If you think about it, this is how humans react when facing the same problem.”

Elder Scrolls IV: Oblivion's advanced "Radial A.I." was heralded for how it simulated the full-day activity of every NPC. Everyone woke up a different times, carried out various chores and jobs, went to particular places to relax at the end of the day, and eventually went back to sleep. Or, on certain nights, went out to the game's fight club.
“Other areas [of AI], such as data mining and evolutionary computation, take novel approaches to exploring large search spaces where an answer to some unanswered question may lie. The goal in this case is often information discovery, going beyond what humans have learned.”
Dr. Courte provides us with one example of data mining, “At convenience stores, a purchase of disposable diapers is frequently accompanied with a purchase of beer. I doubt that most store managers would have on their own determined that placing beer displays near the diapers would be a smart marketing move, but the data, properly analyzed, indicates otherwise.”
Discoveries made by artificial intelligence systems today are more and more complex as the technology develops. How long, then, will it be until we create an AI that is just as, or even smarter, than you and I? “Centuries and centuries,” thinks Kripalani. “This is an amazingly complex problem that involves issues of cognitive models, perception of emotion, language and a myriad of other issues... Do you think we have the ability to teach a computer what it means if someone does not complete a sentence? Everything's so situational it is mind-boggling to try to capture all of the nuances. I think we are a very long way off.”
As extreme as Kripalani’s stance was, Dr. Courte seems to agree. “I’m sure you can find articles in the popular press that predict that soon we’ll communicate with computers as if they were human, much as in Star Trek or other fictional worlds. But you can go back at least three decades and find similar predictions. So far, ‘soon’ has not come, and merely conversing with a computer that seems human is a long way from true intelligence.” He concludes, “I think there is a lot more to learn about human intellect and the brain before we can even begin to contemplate such systems."
|
Page Listing: 1 | 2 | 3 |
|
You must be logged in to post a comment.

| 9/26/07 | |
| 9/21/07 | |
| 8/15/08 | |
| 7/8/06 | |
| 7/15/08 | |
| 5/21/09 | |
| 5/21/09 | |
| 5/19/09 | |
| 5/18/09 | |
| 5/18/09 | |
| 5/18/09 | |
| 5/18/09 | |
