Watson: A New Valentine’s Day Sweetheart

On February 14, Valentine's Day, {complink 2470|IBM Corp.} debuted “Watson” as a contestant on the popular game show, Jeopardy . Watson is a supercomputer powered by an IBM Power7 server and is programmed to handle a massive number of tasks at rapid speeds to analyze complex language and deliver responses to inquiries such as those posed on Jeopardy .

The Watson software incorporates a number of proprietary technologies for the specialized demands of processing an enormous number of concurrent tasks and data while analyzing information in real time. The computer has the ability to analyze the subtle meanings, irony, and riddles that accompany Jeopardy questions. It develops an answer and assesses its probability of accuracy to determine if a response is warranted. All this is performed at record speeds.

This is definitely more than just number crunching the winning moves on a chess board. The Watson software has to understand language, extract information, do it quickly, and then determine a level of confidence in that response.

Although Jeopardy presents an entertaining example of the technology, the real applications are truly awesome. The technology could be applied in such areas as healthcare, to help accurately diagnose patients. IBM suggests other applications such as improving online self-service help desks, providing tourists and citizens with specific information regarding cities, and much more.

Reminiscent of HAL 9000, the onboard computer in 2001: A Space Odyssey , this is a huge leap into artificial intelligence. AI is one of the critical components that will allow robotics to be a reality in our everyday world and not just on factory floors. A new frontier is upon us.

I suspect Watson is also quite capable of writing a romantic message to win over your Valentine. Yes, this might change the way we interact with our computers, but thank goodness we still need humans to deliver the heartwarming hug.

21 comments on “Watson: A New Valentine’s Day Sweetheart

  1. eemom
    February 15, 2011

    I saw Jeopardy yesterday and I also watched a NOVA special on the development of Watson.  The NOVA program was a especially interesting since it followed Watson's 4-yr development journey.  I was amazed on how they went about creating Watson, and continued to fine tune its capabilities.  Working with Artificial Intelligence technology and agencies, IBM tracked Watson's progress compared with real contestant statistics.  IBM debuted Watson to the Jeopardy Executives one year prior to them accepting as a contender.  Only when IBM and Jeopardy Execs were confident that Watson can compete along with the top ranking contestants did the decision to debut him on the show was made.  Machine learning was the technology that made Watson a true contender.

    At the end of last night (only the first round of jeopardy), Watson was tied with one of the contestants.  I can't wait to watch tonight..

  2. JeffWolf
    February 15, 2011

    A very nice article on Watson and his benchmarking against our best and brightest humans, Joanne!

    Watching Watson take the early lead in Round 1 showed us that we truly are at the beginning of a new frontier where computers can perform complex tasks on par with humans. However, seeing Brad Rutter close the gap, as the questions got more difficult, and finish tied with Watson demonstrated the brilliantly competitive spirit unique to humans.

    Since it was Valentine’s Day, I was amused about Watson’s human failings, like not listening to Ken Jennings. No doubt this human oversight in his programming will be addressed in future software upgrades, in much the same way we may need reminders about the importance of careful listening as we go through life.

    With semiconductor technology in development at IBM and elsewhere we may soon see future Watsons approach human form factor and be mobile enough to travel to the Jeopardy! studio. Advances in new semiconductor processes and 3D stacking technologies promise to enable dramatic improvements in system performance, power efficiency and system density.

    Then there is the software. Watson’s large physical footprint is driven in part by human programmers’ inability to productively develop software that exploits fine-grain parallelism, similar to the processing in the human brain. Although fine-grained parallel processing is made available by the hardware, Watson’s software is partitioned using “embarassingly parallel” constructs, using single-threaded programs that are more quickly developed, debugged and understood. Getting Watson to fit behind that podium might require the leap to productive, highly parallel software development from the threshold we humans have been on nearly two decades. Maybe Watson can help us solve this problem.

  3. jbond
    February 15, 2011

    It is going to be interesting to see what happens on Double and Final Jeopardy. Watson is a huge step forward in AI. Even without progressing, Watson's current state could be used for so many different applications and to help so many people. As technology continues to advance, so will Watson. A major step will be reducing the size of Watson. Could you imagine if they could have those AI capabilities installed in ASIMO (Honda's Robot)? We would be seeing our sci-fi movies come to life. 

  4. DataCrunch
    February 15, 2011

    Watson crushed the humans tonight in Double Jeopardy, although it lost the Final Jeopardy question.  I am very impressed by Watson.  Well done IBM.

  5. Tim Votapka
    February 15, 2011

    Great PR for IBM. Good entertainment for the Jeopardy audience. Good work to Ferrucci and his team. Now if you can just make this cool enough to get more youngsters involved in the science of it all, you've got a good, long-term thing going. Dean Kamen might be interested in hearing from you! He's the one who pioneered the organization FIRST (For Inspiration and Recognition of Science and Technology). For anyone who doesn't know what that is; it's a nationwide movement to get high school youngsters engaged and enthused about engineering, high-tech, science and so forth. The idea, run a series of gladiator style robotics contests around the country and make it cool to be a techy; as cool as NCAA basketball. One of the primary goals of the program overall is to get local regions and communities back on the map where technology, engineering and manufacturing are concerned.  The Watson project – like FIRST – could be extremely useful in this purpose.

  6. Tim Votapka
    February 15, 2011

    There'll be many comparisons to Watson and HAL to be sure, but one thing I noticed during a video clip of the project was a little creepy…Dr. David Ferrucci bore a striking resemblance to Dr. Chandra (HAL 9000 creator) in the film 2010. I spotted a parental affinity in Ferrucci's eyes during a segment of a documentary showing Watson's earlier Jeopardy run-throughs. Can't say I blame him. It was just creepy how life imitated art.

  7. Parser
    February 16, 2011

    You are right Watson will miss on heartwarming hugs. I am surprised that it does not carry a female name. It would then bridge the hugs even between women, but let me get serious here. It is definitely good news seeing that Chinese build the fastest super computer in the world. This was our niche for almost a half of a century and it should remain such. We really have to do basic research work and create something new like Watson to broaden our computing knowledge and set the paste to the world. AI has been neglected for some time and replaced by a subset the Fuzzy Logic in almost all consumer products. AI is our true future and I hope that others will follow Watson and IBM will establish main roles in healthcare, medical diagnose and help desks.

  8. Adeniji Kayode
    February 16, 2011

    This is really interesting and really love can only be expressed by a living soul and not a machine,electronic or softwares.

  9. bolaji ojo
    February 16, 2011

    Eemom, Could you please send the link for the Nova special. E-mail it to me. Thanks.

  10. eemom
    February 16, 2011

    I had read that there is some time lag built in before the contestants can buzz in.  They built the extra time in since contestants can ring in prior to Alex finishing reading the sentence so this is to give Watson a level playing field.

    As I watched last night, I noted that if Watson knew the answer with high probability, he was able to ring in first.  I saw frustration on the faces of the other contestants who were trying to buzz in but couldn't.  The only times they were able to buzz in, are the ones where Watson's probability calculations weren't high.

    In order to make this fair for the contestants as well as Watson, I believe this glitch needs to be fixed.  There is a time lag for a human hand to ring a buzzer vs. a computer ringing in, that looked like it was instantaneous.

  11. eemom
    February 16, 2011

    For those who would like to watch the NOVA special, here is the link


  12. Tim Votapka
    February 16, 2011

    And if anyone's interested in a brief retrospect on the AI issue, check out the essay on NPR by Adam Frank:

  13. Joanne Itow
    February 16, 2011

    Tvotapka ,

    I'm so glad you mentioned Dean Kamen.  He is actually going to be the keynote speaker at our SEMICO SUMMIT to be held in Phoenix, May 1-3, 2011.  I agree, efforts such as FIRST and the publicity from Watson will hopefully make math and science 'cool'. 


  14. Tim Votapka
    February 16, 2011

    Well done for having Dean Kamen at your event! His FIRST program has grown at an incredible rate. Long Island's regional competition is now in its 12th year and the energy level these programs generate is incredible. I'm sure your audience will enjoy the candor and style with which Dean presents.

  15. SP
    February 17, 2011

    With the advancement in technology we should really feel lucky to have human to get a hug or talk to. Imagine the era when we have robotos running around and if you feel sad they will come and share a joke or when happy they will say lets party…IBM has been doing great in research and developement. Its great to know about Watson and where cann it be utilized.

  16. Himanshugupta
    February 17, 2011

    eemom, this is quite an interesting and inside information. The time a human hand takes to push the buzzer can be quite significant. In order to take into account, simple statistics can be performed and time lag can be built-in to the computer's response. 

  17. Mr. Roques
    February 17, 2011

    Interesting analysis, maybe they can model the reaction time and the time needed to move the hand to push the buzzer.

    How fast can it learn new things? Or he analyzes things in real time?… I've been following a CMU machine that is learning and “categorizing” ideas and tweeting about it (if you find that something is wrong, you can send a message).

    Here's the link to NELL's twitter:!/cmunell

  18. eemom
    February 18, 2011

    If you go back and watch both Jeopardy shows, you'll see some frustration on the contestant's faces as they were buzzing in but Watson had the “upper hand” so to speak.  I am sure the time lag can be built in and maybe future versions will have that, just as time lag was built in to disallow contestants to buzz in prior to the entire question being read.

  19. saranyatil
    February 22, 2011

    I am just wondering how many different algorithms have gone into it, hats off to IBM for creating such a smart machine.

  20. Mydesign
    February 22, 2011

        Saranyatil, since am working with super computers for the last couple of years, I didn’t felt anything special or wonder about his magic, because he is using a super computer for programming (Supercomputer powered by an IBM Power 7 server).

       A supercomputer is a computer which performs at a very high speed, far above that of normal computers. The primary use for supercomputers is in scientific computing, which requires high-powered computers to perform complex calculations, rendering complex formulas, and performing other tasks which require a formidable amount of computer power. Some supercomputers have also been designed for very specific functions like cracking codes and playing chess. It is commonly using for highly calculation-intensive tasks such as problems involving quantum physics, weather forecasting, climate research, molecular modeling (computing the structures and properties of chemical compounds, biological macromolecules, polymers, and crystals), physical simulations (such as simulation of airplanes in wind tunnels, simulation of the detonation of nuclear weapons, and research into nuclear fusion).

         In my opinion, when compare with very complicated and soficated algorithms for above mentioned tasks, running “a massive number of tasks at rapid speeds to analyze complex language and deliver responses to inquiries such as those posed on Jeopardy” are quiet simple.

    There are 2 types of computing

    • Capability computing is typically using the maximum computing power to solve a large problem in the shortest amount of time. Often a capability system is able to solve a problem of a size or complexity that no other computer can.
    • Capacity computing is typically thought of as using efficient cost-effective computing power to solve somewhat large problems or many small problems or to prepare for a run on a capability system.
  21. saranyatil
    February 22, 2011


    thank you for your clear cut view & information you shared.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.