My Inner Nerd Gets Excited: IBM’s Watson Wins Jeopardy

If you are not into computers, this topic might put you to sleep. 🙂

From Valentine’s Day, Monday February 14, until Wednesday February 16, 2011, two former Jeopardy Champions Ken Jennings and Brad Rutter battled a computer. And not just any computer. It was a machine hundreds of people from IBM and several prestigious colleges across the country worked on for over four years in order to create a Jeopardy Champion.

In hindsight, it sounds almost ridiculous that a massive corporation listed on the NYSE (as it’s still known of as for now) would devote over $30 million for a chance to have its software, hardware, staff and reputation devoted to one challenge: beat humans on a televised game show.

It seemed simple, Watson was going to perform the same feat as Google did for millions, right? You enter a query and it’ll return the first hit. However, the difference is massive. Watson never accesses the internet. It doesn’t phone home. It is a standalone machine. All the data it needs is stored in its database. Practically the entire Library of Congress is stored inside its hardware. That is millions of books. Millions and millions of words. Google finds data based on page ranking, web sites with the most links which determines popularity, and other unknown algorithms.

Plus, Google doesn’t respond with the nice voice that Watson has.

Yeah, that’s one of the key differences with this machine: Watson speaks! He, er, it asks for the next clue when its turn is up! It even says, Please!

Watching the show for the three days it aired, all I thought of was inquisitive HAL from the movie 2001, the Cyborg from the first Terminator movie (that scared the daylights out of me), the fussy sounding, talking car named Kitt from the television show Nightrider, and the informative, perfunctory computer from Star Trek (the original) and Star Trek: TNG.

Watson was able to understand the near intuitive based quirk of a Jeopardy “answer” in order to submit a “question”. It understands natural (spoken) language and the nature of a riddle. I read that the machine used to take two hours to process a question eventually getting down to under three seconds, because it can learn.

After Watson calculated an answer, he ranked it, indicating a level of self-confidence. If it was over a threshold like 50%, Watson pressed the buzzer. The machine also wagered for the last round – the final Jeopardy question.

It received all questions via text, but in the future Watson will be able to listen and probably see. That’s creepy.

I celebrate the technological advance, but I’m also freaked out by it.

I wasn’t surprised to see HAL, er, Watson win. He spanked both men with $77147, much higher than the combined winnings of $24,000 (Jennings) and $21,600 (Ruttner). However, no one went home empty handed. Watson received $1 million, all of which was donated to charity. The two men split their winnings with charity 50-50: Jennings won $300,000, and Rutter won $200,000. I liked the joke Ken Jennings offered during the final question, “I, for one, welcome our computer overlords.”

I don’t know if I do. However, I do appreciate this technological breakthrough in computing power. In a few years, we may all have our own personal Watson on our smart phones, smart pads, and whatever other technological tools come our way. The danger and the reward in this new technology, is that the more it helps us think, solve problems and delves deeper into complex situations – the less human beings and our intuitions are required.

With one hand technology giveth, and with the other it taketh.

I suppose the only way to stay ahead is to adapt, find the areas where we can offer assistance, be innovative, and continue to do things in areas where technology cannot take our place: creativity and ingenuity. Well, at least for now.

Let’s hope they keep Watson from becoming self-aware, emotional … and away from the military. Especially predator drones. 🙂

Geek Notes: I found out it was mostly programmed in Java and C++. Wow. I had Java and a bit of C++ in school. Those aren’t easy languages to understand. Maybe I’ll take another shot at Java one day.

Update: Video Clip – How Watson Works by Dr. Ferrucci of IBM. Mentions the compter from Star Trek: TNG, which he wanted to emulate.

Share