IBM's Watson
-- posted December 2014 --
This blog is about one of the most sophisticated artificial intelligence machines of the present time – Watson, developed by IBM. Namely, Watson participated in the Jeopardy quiz show, and won $1 million in a competition with two of the most successful contestants in the history of the quiz.
Here is a link to a video from the competition:
https://www.youtube.com/watch?v=P18EdAKuC1U
Watson uses access to databases (books, encyclopedias, dictionaries) to search for answers to the quiz questions. What makes Watson intelligent is its ability to ‘understand’ the questions. If you have ever watched Jeopardy, you would know that the clues are given in a puzzling and nuanced way, so that even for us it is often difficult to quickly tell what the answer is, even if we knew the answer.
To be competitive, Watson had to have capabilities to decode the witty wordplay of the questions, search through hundreds of millions of documents, and find the answers in a matter of seconds. When providing its answers to a question, the three most probable answers by Watson were displayed on the screen, along with the ranked percentage of its confidence in the answers it comes up with. Although the machine outperformed the human contestants, Watson made a few mistakes on very simple questions (I mean simple for us).
The following 7:53 minutes video explains the computational processes behind Watson’s intelligence.
https://www.youtube.com/watch?v=U6rvaWaiZNg
Currently, Watson is being trained to help doctors as a clinical decision support system. What it means is that Watson will employ access to vast medical data, and in consideration of patients' symptoms and medical history, the machine will provide diagnosis and treatment recommendations.
