Watson's marketable skills

Jeopardy was fun, but Watson's practical applications are what's really interesting.

watson2.pngArtificial intelligence certainly isn’t a new concept, but as some have pointed out in the days since IBM’s Watson computer skunked two expert trivia contestants on Jeopardy, there just wasn’t any money in it. As the digital age advances, however, intelligent search is becoming a marketable product.

In a recent blog post, Stephen Wolfram explained the differences in Wolfram|Alpha and Watson search models and how each approaches data. He described Watson’s search structure:

The IBM Jeopardy approach has to do with answering questions from unstructured textual data — with such potential applications as mining medical documents or patents, or doing ediscovery in litigation.

The immediate marketable application of Watson indeed will be in the healthcare industry where the unstructured search capability will be combined with speech recognition technology from Nuance.

Others have pointed out applications in gaming, and IBM has pointed to online helpdesk and tourism applications. Digital Power Capital’s Mark Mills declared Watson the beginning of the next “great cycle in computing.” But referencing Trevor Pinch’s post, Mills said the technology isn’t quite there yet.

tags: , , ,
  • http://www.i-programmer.info Dr Mike James

    As I point out in the article at:

    http://www.i-programmer.info/programming/artificial-intelligence/2012-watson-wins-jeopardy-trick-or-triumph-.html

    There may not be as much to comercialise about watson as IBM might think.
    The format of Jeopardy is such that it makes the statistical approach work better than in “simple” question and answer situations.
    Jeopardy provides lots of information in the clue which can be used to pin down the entity being described by the question.

  • http://www.allianceglobalservices.com David Rader

    There’s an interesting webinar coming up about how LexisNexis uses semantic technologies and NLP to improve their Legal search results. Interesting problem across 5 billion documents!
    http://bit.ly/gS97Uc