This recent Wired article on AI risk was rather ok: http://www.wired.co.uk/news/archive/2012-05/17/the-dangers-of-an-ai-smarter-than-us -- Aleksei Riikonen - http://www.iki.fi/aleksei