October 11, 2017
One of the weirder areas in cybernetics is the area of artificial intelligence. As researchers and programmers have sought to replicate the way that the brain works, a number of interesting oddities have come up as the logical realm of computers meets the sometimes irrational realm of the human mind. However, this is nowhere more obvious than in the realm of gaming, where the ability of a non-player character to make its own decisions while following the directives of story and programming
While most of the characterization is no doubt due the voice acting, the characters still need to behave correctly during play, this is where having intelligent characters. Originally one of the problems with character behavior is that characters tended to take their directions extremely literally, which led to a number of problems where characters committed de facto suicide; it was not uncommon for the characters to follow directions to “go that way” right over a cliff. Even otherwise intelligent characters would engage in monsters in combat sometimes even before they could receive instructions from the player. Suffice to say that this caused innumerable problems.
Handling these problems actually ended up helping in other fields. The “go that way” problem was solved by establishing parameter where the computerized characters had limits on where they could go, and they got better at stopping before encountering danger. This ability was translated to electronic explorers, giving them a chance at survival as they had to debate their odds of survival. Figuring out how to make characters obey commands better allowed better use of intelligence to be used by battlefield robots to better follow the squads they were attached to and enabled them to help in the defense of those squads.
One of the most interesting developments was actually in chess. A chess computer in 1983 apparently began cheating and the programmers could not figure out the problem. Then, they had a chess player look at the recorded games, and they were able to ascertain that the first person to cheat was a person, possibly a player making a move in error. In order to save memory the programmers had neglected to install a move-checking sub-routine on the assumption that the grandmasters they were using would never cheat or make a wrong move. A wrong move was made and the program caught it, and learned from it. It was soon making mistakes on purpose, which quickly caused some of the grandmasters to complain about it cheating. Installing a move-checker stopped the cheating.
It has been used in other games for other functions. The Anki racing system establishes a course for the cars that they must stick to in order to win. The system uses a relatively simple piece of software to adjust the speed of the racers so as to ensure that they can handle the turns and overtake the lead vehicle when possible. The software even simulates cars following each other even though there is no real advantage at that scale. Artificial intelligence has been able to be used in games to make them more challenging, and in some cases has been in used in other fields, making life both easier and more fun for all involved.
September 11, 2017