Some researchers say that the best test of artificial intelligence is to have a human interact with the program and other living humans, then have the first person try to guess which one is real and which one is not. I happened across this article and it made me wonder if automated bots on the internet are getting intelligent enough to pass this test. Ironically, these bots were designed to play games and not to pass the aforementioned test.
http://news.bbc.co.uk/2/hi/technology/4165880.stm Student held over online mugging
Police in Japan have arrested a Chinese student over the use of a network of software "bots" to steal items in an online role playing game (RPG).
Players were attacked in the game, Lineage II, and their items were then sold for cash on auction sites.
The attacks were carried out using automated bots, which are difficult for human game players to defeat.
The student, who was abroad on an exchange program, was arrested in the Kagawa prefecture of southern Japan.
In Japan, as in England, there are no specific laws to govern trade in virtual possessions.
Bot traps
Use of bots is a frequent problem in online gaming, and most game publishers have invested heavily in trying to eliminate them from their games.Bots appear in games in the same way that human players do, so there is no easy way to tell which players of a game are not real.
Instead, complex techniques called bot traps have to be used to trick bots into revealing themselves.
For example, in a first person shooter (FPS) game, players who seem to be moving too fast or pinpointing a particular point very accurately raise alarm bells.
Asking direct questions or placing players in unusual situations in the game are techniques which are often used by administrators to identify bots.
However, for every improvement in bot detection, the bots themselves become more complex and more difficult to spot.