Robot... or Human?

For a machine to pass the Turing Test -- that is, for a robot to act indistinguishably from a human -- we would need to program it to behave with some of the more complex aspects of human personality. I think that in order to accomplish such a task, the computer would need to "study" hours upon hours of humans chatting with actual humans. This way, it would learn some common vernacular, possibly learn meanings of word-sets beyond the literal, and be able to memorize some standard responses to certain questions. Also, it would need to problem solve; in case the person on the other end of the chat made a typographical error, the robot would definitely need to be able to figure out what the word was supposed to be, dependent on context. It would need to have some knowledge of emotional expressions, an encyclopedia of references, and basically either be able to discuss anything a human wanted to discuss, or express in a human-like way that it knows little of the topic. The computer would also need to be able to learn from its own mistakes, and in case of a mistake, understand the correct protocol for recovery. Basically, the Turing Test doesn't test for real intelligence, the way most people think of it. It could be passed by a computer with a large memory capacity and extraordinary ability for brute memorization.