By now you’ve heard the news:
IBM super computer Watson came away victorious during Jeopardy Wednesday, but not before the game show’s former champions Ken Jennings and Brad Rutter rallied a formidable defense. In the end, however, the humans were no match for Watson, which won with a commanding lead of $77,147 after three days of Jeopardy play.
Watson is 16 TBs of memory, 10 racks of commercially available IBM Power 750 Servers, and is stuffed with all kinds of facts from history, pop culture, science, dictionaries, bibles, and movie databases. Its software’s challenge was to work through Jeopardy’s intricate pun-driven wordplay to produce the right answer (or question) before the two champions. Compare that to a Google or Bing search, where strip down your query to the simplest form – and still get pages of unrelated content.
The heart of Watson is DeepQA, software developed to do answer an open-domain question – that is, to answer a question the way a person would ask it. Right before our eyes, Watson learned from its mistakes. By “listening” to another player’s correct answers, it adjusted within a category and put itself back on the winning track.
Watson will change the way the general public views what computers can do, and its capabilities are bound to create jobs for programmers. Although I only found one job post relating to the open source backbone of Watson’s Unstructured Information Management Architecture (UIMA), the listing from IBM actually seeks multiple people: The software giant is “investing heavily in the hardware, software and machine learning infrastructures necessary to scale this research to the next level.”
A broader search on “semantic analysis” produced four listings including Machine Learning Software Engineer/Java Engineer. A search on “machine learning” produced 146 results, and a search on “artificial intelligence” found 49 listings.
This is the ground floor of a new area. Although it’s now spearheaded by IBM, businesses whose products manage vast amounts of data are bound to jump in as Watson goes prime time.
— Dino Londis