Meet Watson. Yes, those IBM researchers are at it again, with their crazy attempts to ensure the redundancy and probable destruction of the human race. Or, more accurately, another attempt at a human-beating game-playing computer. This time the goal is a machine that can best the top human contestants at US gameshow Jeopardy! next month.

Unlike its chess-playing predecessor Deep Blue — where the problem to be solved was largely computational — DeepQA/Watson’s ability to answer trivia questions depends not only on swift information retrieval but upon some serious advances in natural language processing. In short, the key hurdle that has so far limited my dream of a future house/virtual butler to which I can talk. So imagine my surprise when, in yesterday’s test run, Watson actually triumphed.