"Smartest Computer on Earth"

Current State of General Topic Area

Claiming that another computer is comparable to Watson is absurd, it is nonexistent. The other computers that have any similar features are Watson’s “parents” like Blue Gene or Deep Blue. Some of Watson was made from the information and the development of past supercomputers.

            Deep Blue was able to beat the world champion in chess due sync development. That meant a programmer would sit at the computer and literally type rules, line after line, into the computer. The sync list generally has 600 million rules listed. Most of the rules are common sense to humans (“Will Watson Win Jeopardy”). However since computers can’t think conceptually, there is no such thing as common sense in computer vocabulary and everything needs to be explained.

            One of the main goals of developing Watson was to have the computer learn from examples and mistakes much like humans do. This idea is called machine learning (Flach). Watson learns from examples and patterns given to him. For example, a programmer will show Watson the letter A. Well there are many different types of A’s and how people write them. So in order for Watson to understand the letter A, several different examples of A’s are processed into the computer. Eventually Watson finds a pattern and learns the letter A and can recognize it (“Will Watson Win Jeopardy”).

            To prepare Watson for Jeopardy, the programmers at IBM actually added a cheat sheet for him. They downloaded all of the Jeopardy questions they could find and provided the answers. If one has ever watched Jeopardy, it can be seen that one the categories are broad and that also the questions have puns and word play. Humans understand word play, past computers didn’t. So Watson, through machine learning, studied all the questions and developed a pattern. This allows Watson to develop three top answers within milliseconds and analyze which is the best answer.   

            Here is a great example of how past computers answered questions. Blue Gene was asked this question from a list of Jeopardy questions,

            “In 1698, this comet discoverer took a ship called the paramour pink on the first purely scientific sea voyage”

            Blue Gene would answer Peter Sellers, who is the star of the Pink Panther movies.

Absolutely wrong, but how did a computer get it so wrong was the next question.

Here is how, Blue Gene analyses millions of documents and zoned in on key words such as pink, paramour, and sea. There was a Pink Panther movie in 1963 that involved a ship and a character described as a “Paramour.” That document generated a lot of hits with the key words and Blue Gene chose the star of the Pink Panther Movies, Peter Sellers (“The Pink Panther”).

            Watson with his machine learning and understanding of word play was able to answer correctly with Edmond Halley. When Watson answered this question correctly and within milliseconds, it was a huge celebration for the IBM research team. This proved that their vision of a human brain computer was possible (“Smartest Machine on Earth”).