Thursday, February 17, 2011

The Fall of Man

Watson, everyone's new favorite super-computer, is a cheater! I just wanted to get that off my chest. What are the odds that Watson's very first box selection (the 3rd of the game, iirc) was the daily double? Well that's exactly how it played out and that's when I knew this was not going to be a victory for human-kind, though I suppose technically man created Watson so any outcome would be ruled favorable. It is remarkable to see how well IBM has done creating a Jeopardy Juggernaut, but even more amazing is how far they have to go. Watson does not have a microphone and has no way of reacting to incorrect human responses. And even if he did, his algorithms are designed as such to create only one "real" answer. I noticed on many of the art questions that the secondary and tertiary responses were source information such as the name of the artist or museum of the work, which wouldn't help in this situation.

I also thought Watson's change of strategy between rounds, from navigating the board horizontally across categories to a more random selection before settling into a vertical approach in the second round was peculiar and a bit transparent. The super-computer was clearly searching for the daily doubles, found them successfully and then went to auto-pilot with a comfortable lead. Why does the computer take its foot off the gas pedal? The trick to stumping Watson is clearly in our complex understanding of the language, such as referring to passages in books to identify the character or through the association of two seemingly separate facts. I would also be interested in how it would compete in the more abstract word-play categories like anagrams, crossword puzzle clues and poetry/literary references. All in all though, I thought Jeopardy was a great platform to debut the next step in artificial intelligence and enjoyed a show that I have not watched in some time.

I didn't get to see the last episode of man vs. machine, (same old same old i gathered) because I was enjoying Brooklyn Brewery's East India Pale Ale with my in-laws. This name conjures up thoughts of super-hop and bitterness that should make your face pucker, but i found this brew didn't live up to it's name at all. It tastes much lighter than other IPAs, dominated with pineapple-orange citrus undertones and a sweet, mild hoppiness with just a bitter hint (pine??) at the end. The aroma was nothing special, and was the first signal I wasn't going to get what I expected... a slap in the face!! I wouldn't turn this brew down, but now i know what to expect (and not expect) from Brooklyn's tame offering. I would suggest a change in the name though to avoid future confusion, perhaps defining it as an American Ale or a Not-so-Pale Ale would be better suited.

No comments: