Artificial Intelligence Thoughts: Turing Test

I was just reading the first chapter of John Battelle’s “The Search” and he proposes a future where artificial intelligence is realized as an extension of search; coming from the breakthroughs in web search. It would be simple create a bot that would search the Internet for answers to any questions posed to it in a Turing Test (see definition). And if the answers happen to come out in a natural manner the tester could be fooled into thinking the bot is a human. Of course, this doesn’t really solve the problem of AI. It passes the Turing test, but it isn’t actually intelligent. Though it can learn from humans, it wouldn’t know which answers available on the Internet are correct and which aren’t.

My other idea for passing the Turing test is to create an AI wiki. Where people submit information in a specific format (noun + verb for example). The AI bot then uses NLP, Natural Language Processing, to parse a given question and look up the corresponding answer in the wiki database. The wiki angle would allow the populations of the world to contribute to the bot’s wealth of knowledge. This would be more precise than using the entire Internet as a database, but it would take time and effort to be able to answer any question imaginable in any language. Is anyone already doing this?