Hello hello! The Editor and I are pretty fresh off our trip to the EMPAC center at RPI, and the super awesome lecture about the Watson super computer from IBM. It was pretty much the most awesome lecture I've ever sat in on. We showed up about 6:15 or so, grabbed some free food and drinks (most awesome part about a lecture on a college campus), and were in our seats for the lecture at 6:45. The lecture consisted of three panel members; Dr. Chris Welty (who is completely and totally smart, well spoken, and funny), Dr. Jim Hendler and Dr. James Meyers. The Editor referred to these guys as "the Mystery Science 3,000 of Computers".
Chris Welty is an RPI grad, and fun fact! It took him 14 years to graduate from RPI. He still has the longest standing record for most semesters attended. Dr. Welty did the bulk of the talking, and had an amazing PowerPoint presentation the he showed, that explained Watson in a way that a non-genius (like myself) could understand. Unfortunately, I couldn't find the presentation anywhere online. :-(
Below is a short video giving a very basic overview of what Watson is and how it works. I'll have The Editor write up a longer, more detailed explanation later in the week about the lectures and Watson, as he has a MUCH better grasp on the whole thing than I do. I will be giving you the dumbed down explanation.
Watson is basically a server system storing tons and tons of information. Watson is given a question and because of all the servers, can come up with an answer pretty quickly. If you watched last night, the screen showed three possible answers that Watson came up with, and whichever one had the highest percentage correct, he rang in for the answer. One thing we learned last night is that Watson will NOT buzz in unless there is a greater that 80% probability that the answer is correct.
As you witnessed last night, Watson is not perfect. Sometimes the computer does not understand the question the way a human brain would, and gives a wrong answer. Also, since Watson is incapable of hearing, or seeing, he buzzed in with the same answer that Ken Jennings incorrectly answered right before he did.
Watson's "avatar" was pretty cool too. The blue and green "wires" become more active when Watson is "thinking", and they settle to the bottom and turn red if he gets an incorrect answer. It's pretty amazing.
With the technology that IBM has created with Watson, there are a ton of other projects that they are working on, to make lives easier. Most notable is the writing of a medical program which doctors can use to better diagnose patients. Basically, the doctor would input the symptoms, and match them up against other documents and studies from other doctors, colleges, medical journals, etc. around the world to get an answer SPECIFIC to the question they asked.
"Why couldn't they just use Google?" you may ask. Google is a great search engine if you want a broad answer on a simple question. However, the more complex the question becomes, the less information Google can give you. Generally speaking, it will only give you the most popular answers, and that may not be EXACTLY what you're looking for.
Unlike traditional search engines, Watson is capable of "thinking" more logically, so it can better understand what it is specifically that you are looking for. However, as previously stated, Watson is not yet perfect, and you wouldn't want an imperfect computer diagnosing you with a life threatening illness, if you don't have one.
I'm not even a huge tech dork, and I find Watson absolutely fascinating. The Editor (who is a computer engineer) referred to the Watson project as "inspiring". The AI that Watson uses can truly help to make the world a better place, once it is fine-tuned, and ready for everyday applications. Let's just hope it doesn't become self-aware. ;-)
I'll be back with more info tomorrow! I expect that you are all watching Watson this week, and if you're not, you should absolutely look up the episodes and give them a look!
For more information on Dr Chris Welty click HERE
For more information about the Watson project click HERE
BUT WAIT! THERE'S MORE!!
Here are The Editor's notes on Watson day one:
Watson Day 1 Notes
Watson does not hear or see
As such, there are no video or audio answers. Jeopardy! eliminates these when
there is a blind or deaf contestant, so no problem here
Watson is actually "fed" a text file with the clue once Alex begins reading
(This miffs me a bit, they couldn't add a camera with OCR to read the clue?)
Watson is not connected to the internet
Since game shows are federally regulated now (and have been since that big scandal in the 50's)
there were two auditors at the taping, ensuring all the rules were followed.
The team couldn't have laptops, internet connections, etc.
(I did note that Alex specifically says that the game is an 'exhibition' a few times though)
One of the best reasons to attend the screening at EMPAC are the commercial breaks (though free chicken tenders is a close second). During the breaks the commercials get muted and the panelists make comments and observations, like a less-funny more-geeky MST3000!
You'll notice during day one that for a "Decade" category question, Ken buzzed in and incorrectly answersed "the 20's". Watson then buzzed in and answered "The 1920's", which prompted Jennings to turn and stare in disbelief. At the next commercial break, Chris explained that they had pondered the scenario where a contestant answers incorrectly, Watson should remove that from its possible answers. They didn't implement this since their statistical analysis showed it was very very unlikely to occur, Oops! ;-)
Stay tuned for more coverage and notes from Day 2!
No comments:
Post a Comment