Kacie  & Melissa

Watson is an Artificial Intelligence that was made to compete on the television show Jeopardy! It has a large memory with much information stored in it. It was programmed in Java and C++. It analyzes language, searches through its memory to find what it thinks is the right answer, and chooses the answer that it thinks is best.

IBm had just finished triumphing with their chess project Deep Blue. In 2004, they decided to create a supercomputer that could beat Jeopardy! champions. They started with the computer Piquant, but it was only right 35% of the time, and it took several minutes to answer each question. Then they moved on to Watson – the first time they tested it, it only got 15% of the questions right. In 2007, IBM received full funding for the project  and an entire staff to work on it. In 2008, it was ready to compete, and did so in 2010.

Watson’s general application is just answering questions. Its main feat was learning to analyze the English language and its slang.

One response to “Watson”

  1. Lekan

    Yes. Learning English colloquialisms and understanding took a great deal of research to do correctly, eventually using many complex probabilistic models to capture the ease with which we humans can not only understand language, but quickly connect it with disparate other pieces of knowledge scattered in our brains.

    Besides just answering questions on Jeopardy, Watson is being developed as a system that may be able to aid physicians in diagnosing patients. So far, it has been quite successful.

Leave a Reply

You must be logged in to post a comment.