Artificial Intelligence – Taking over Jeopardy?!
Will super-computers one day take over Jeopardy full time? Bots competing against other bots for money?? Probably not… but this week the classic game show, Jeopardy is featuring a powerful super-computer named ‘Watson’ who will compete against past Jeopardy champions to see if this computer can outwit the some of the show’s best human contestants. One name you may recognize is Ken Jennings who holds the all time record for consecutive wins at 74 games in a row in 2004. The computer, Watson, has a lot of people talking and many people are interested to see what he can do.
A little bit of history about Watson and the people who made him: IBM is the creator of Watson the super-computer who runs on about 3,000 parallel cores and 16 terabytes of ram which is the equivalent to 6,000 high-end at home computers. One point that surprised me about Watson is that there is no access to the internet allowed. Watson can only search for answers that are already stored in its own memory (storage). For this, creators made sure that Watson was loaded up with tons of encyclopedias, Wikipedia entries, NY Times issues, the Bible, and many other sources of information that could be asked on the show.
The whole idea for the project that gave life to Watson was based on the computer Deep Blue who challenged world champion chess player, Garry Kasparov, to a six-game match back in 1997. IBM had built this computer and that eventually won 2-1 games with 3 ties. The difference and challenges that lay in the path of constructing Watson vary tremendously compared to Deep Blue. Deep Blue was a computer that used based logic of all possible moves that its opponent could make throughout the match and then moved accordingly. The challenge with Watson was that, Watson needed to be able to understand human language, puns, wordplay, and all of the complexities of metaphors that we have in English. It took 25 programmers and researchers, 4 years to put together all of the algorithms that made it possible for Watson to search through massive amounts of data to find an answer that it was confident enough to share with the Jeopardy contestants. There is a whole process that Watson passes all of the data through so that it can finally weigh what options are best suited for the answer. When Watson has a high enough probability of being right he says the answer in his very monotone voice.
One of the coolest parts that I learned about Watson is that with the algorithms that are used, Watson is able to teach himself new information and how to understand what is being communicated to him. When he gets answers right he gains confidence and then learns where he succeeded (or failed if he was wrong).
So at this point you may be wondering, why has IBM wasted so much time on money on a machine that was built to beat humans at in Jeopardy?? Well…
Watson is a great example of how technology could one day help in many different professions. Although Watson would be a device that would be too expensive to build for a small group of people, Watson has given light to greater artificial intelligence that can be used for the betterment of society. Think if all of the doctors in the world had a Watson to go to for answers about a specific person, at a certain age in his or her life with a unique medical history in their family. Watson could theoretically weigh all of the different options for a diagnosis and/or solution to their problem and then give statistics on how likely each answer is. I’m sure that from this you can imagine the endless possibilities that a machine like Watson is capable of solving.
You can all watch Watson in action this week on Jeopardy: Monday, Tuesday, and Wednesday (2/14, 2/15, 2/16) on CBS — the local channel is WMBD, channel 7 or 31.1– at 4:30 pm.
Here are some of the sites that I used for information about Watson, and if you did a simple Google search you should be able to find a ton of other sites as well… Watson on Jeoparty and a Youtube clip.
Don’t forget to take the quiz!
Entry filed under: Uncategorized.