February 22, 2011 – The IBM “supercomputer”, Watson, left its opponents, arguably the best Jeopardy players of all time, in the dust in last week’s competition. IBM engineers spent four years and dedicated computing power equal to 2800 desktop PCs to the effort. They ended up with a machine that learns as it ingests and “experiences” answers and successful questions (the Jeopardy format).
The Wall Street Journal inspired this discussion in their article, “Can a Computer Win on ‘Jeopardy’?” Watson got a good deal of press, and a full episode of the PBS series, Nova, The Smartest Machine on Earth.
IBM plans to use the knowledge they’ve gained from building a system to interpret natural language, including the metaphors, puns, similes, and other language tricks included in Jeopardy clues, to synthesize answers to medical questions from a wide range of information sources. Sound familiar? Much like what knowledge workers do every day – researching, abstracting, re-combining, synthesizing essential information in new content formats and uses.
There are those who predict that robots will replace humans in high-skill jobs. (Will IBM’s Watson put your job in jeopardy?)
For Jeopardy, Watson’s ten banks of linked servers are attempting to do the job of one human brain. And though victorious, Watson made a number of mistakes. We humans are still the champions of natural language and of making the linguistic connections necessary for understanding each other. With a seemingly inherent skill, we grasp inflection and implication as well as context.
So, for the foreseeable future, we’re still most interested finding the best ways to pair human and machine, taking advantage of what each does best. Access Innovations’ Machine Aided Indexer (M.A.I.) uses rules written by humans, the ones experienced in natural language expression and context, to guide a computer in its speedy parsing of volumes of text. We believe it is the very best way to assign indexing terms to content to make finding it a breeze. And, it costs considerably less than a Watson.
Sponsored by Data Harmony, a unit of Access Innovations, the world leader in indexing and making content findable.