George in Second Life
Rollo Carpenter winner of the 2006 Loebner Prize and the creator of the George artificial intelligence revealed on last night's segment plans to adapt his creations for Second Life. Listeners to the segment (which you can hear here about 15mins in) will have realized that George's conversation is a little odd; we're not quite at the stage where an AI will be passing the Turing Test. Part of the reason for this is that George is in many ways an embodiment of Philosopher John Searle's Chinese Room problem. As I understand it George's utterances are drawn from a vast (and heavily pruned) database of previous conversations. His ability to talk is based upon his understanding of the relationships between words, not about grasping the meaning or words or how they relate to the physical world. Of course giving an AI an awareness of the world is a complex task, the data processing required is enormous. Which is why placing an AI in Second Life is such an interesting experiment because it is a simplified world, which (depending on what Carpenter's plans actually are) an AI can interact with, and in a limited way percieve.