I watched jeopardy for the first time in years. And, yes I use to watch it all the time. I know it’s weird. Usually as you get older you start watching shows like Jeopardy, but I liked watching it as a kid. And this is why:
Anyways, watching IBM’s computer, Watson, compete on Jeopardy was definitely a cool thing to see. While I wasn’t surprised to see it do well, I was impressed with the computer’s ability to quickly come up with potential answers and ranking them in according to relevance and confidence in its answer. When Watson did get the chance to buzz in, it rarely provided an incorrect answer. And, compared to a human, Watson buzzed in for almost every question.
But the question remains, is this ground breaking?
All I could think about during the episode as they explained the back story and the development of Watson was what the impact on search would be. Watson obviously has a very detailed algorithm that is based on searching for relevant information.
I think there is a direct correlation between Watson’s abilities and the future of natural language search. Google is already working on natural language search and already has Google instant. This may be the basis for future algorithm changes that rewards website with high quality content.
Watson’s ability to decipher the meaning of words and its association with other words is unbelievable and seems to be the precursor to a more detailed algorithm for search engines to use to display even more relevant results.
In the few questions that Watson did get wrong, I wondered if his performance or ability to find the most relevant answer was impacted by the type of question that was asked, in a similar way to search results being impact by the type of words and wording used in a search query.
For example, a more general question may be more difficult for Watson to answer because there are so many possible answers. Similarly, a very general search would yield less specific search results.
It will be interesting to see what developments arise out of this experiment!