I for One Welcome Our New Computer Overlords
Ken Jennings has a way of making history and for delivering the unexpected. For a true know-it-all, he comes across as anything but. If there could be an everyman’s master of knowledge, it is he. That he has a personality in sport that doesn’t tend to produce them only adds to his legacy, for the stereotypical quizmaster is more like the character played by Dustin Hoffman in Rainman than one who is personable, endearing, and cleverly witty. If the name Ken Jennings sounds familiar but is hard to place, you will have to go back to mid-2004, and it would have helped were you to watch the occasional episode of Jeopardy. He might have gone unnoticed too had Jeopardy not changed the rules restricting players to a maximum five game win-streak six months prior to Jenning’s arrival.
Before Ken and since Ken, the longest winning streak spanned eight episodes. Ken won more than nine times that amount. He won 74 straight matches. Words like staggering, unbelievable, don’t begin to describe the achievement. As a frame of reference, the person who beat Ken, didn’t go on to win the next day. Winning once is doable. Doing it again and again becomes almost algorithmically harder. Not only did Ken seem to have more knowledge than anyone else, he could recall it quicker, all while under the immense pressure of being broadcast on TV. Even when he didn’t know the question, he seemed to find it. His speed, precision, poise, and sheer dominance all combine to elevate him to the realm of world-class athlete and a true champion. That he has a personality, made him more than a statistic. It made him memorable.
Jennings has been bested twice – during his drive for 75 wins and in the Ultimate Tournament of Champions when pre-rules change undefeated winner Brad Rutter showed his own version of astounding intellect. For Jennings, this week marked his third loss and Brad’s first. Before this week, we would have called Brad’s and Ken’s abilities superhuman. We might consider them individuals with a super-computer for a brain, able to store and recall information tucked into the recesses of the brain. We might have considered them more machine than human. Today, they are definitely human, and that’s what is scary.
The defeat of Ken and Brad happened simultaneously. They both played together again, but unlike before, neither won. This defeat was truly special, monumental, perhaps even life altering. Were this the 1960’s, chances are the entire nation would have witnessed what took place this Monday through Wednesday at the IBM campus outside of New York. The event sounds prosaic at first. It was, as might be expected with the reuniting of Ken and Brad, a version of Jeopardy. That it took place at the IBM campus hints at the distinction. This week, in a special three day event, Ken and Brad, played a machine – Watson. The name does not pay homage to the quieter but seemingly more book smart sidekick to Sherlock Holmes. It pays deference to IBM’s founder, Thomas J. Watson.
In an excerpt of Stephen Baker’s “Final Jeopardy: Man vs. Machine and the Quest to Know Everything” which chronicles Watson and team from birth to victory, we read of an almost apocryphal story about how the Jeopardy challenge came to be. It involved this executive eating dinner during Ken’s streak and watching diners make their way to the bar to watch Jennings. It was there that the idea to pit man versus machine in Jeopardy originated. That IBM chose to do so made sense. The company has a history of using computers to tackle problems that once seemed only reserved for the human brain. They achieved fame in 1997 when Big Blue defeated the world chess champion and legend Garry Kasparov. It was a rematch of a battle that pitted Big Blue’s predecessor, Deep Blue, again Kasparov. As Baker’s book points out, chess holds a special place in our psyche. Unlike trivia where it seems you can study more and get better, chess, speaks to a level of internal abilities, a true complexity of thought outside the realm of study.
Big Blue proved that chess is actually just a game and that a computer can play better than a human can. That the average person might struggle to become proficient or expert is what makes it seem a natural for a computer. That as part of being human we speak, memorize facts, understand cultural references, and parse language for meaning, we don’t consider just how complex it is. The closest thing we have is Google, for seeing what a machine “knows,” or more accurately can retrieve. What IBM did with Watson was even more demanding because it involved both the consumption and retrieval of vast amounts of information. It had to understand the game – what categories to choose, what to bet, and when it should buzz. All much harder than they sound. Going to Google and typing in keywords of many of the questions shows what a great job Google has done, but even that relies on a person to extract the answer. It also requires a deeper understanding of what the question required – a word, a phrase, a proper name, etc., and doing it quickly, correctly, repeatably. Watson did all that.
Watching Watson was at times inspiring but mostly like watching the Judgment Day countdown begin. Watson wasn’t perfect, but the computer was so dominant that its pleasant computer voice only added to feeling of impending doom. You could see a visually frustrated Jennings as his historically lightening quick buzzer skills could not compete with a computer that could press the button with faster speed and a true machine like precision. It wasn’t that Watson knew more answers than the other players, it was that those qualities which make it not human were prominently on display – the type of qualities which you can just see manifesting itself in the future when we try to deal with Watson instead of people.
Much has already been written about the technologies underpinning Waston and their applications in the “real world.” A great deal of speculation exists about how to best leverage Watson in the field – a machine that can both do a better job than Google at answering, “Who has the second longest streak in Jeopardy” to suggesting alternate treatments to physicians based on the world’s knowledge of medicine. As far as the show goes, Watson’s only flaw is that he/it is made by men, and almost all of its mistakes on Jeopardy could be traced to human error. But, the human error making it vulnerable was so far exceeded by all that made it so invincible that we’ll take Judgment Day as opposed to the benign computer in Star Trek. Watson came across as more than a machine. It came across as a weapon not a tool, and that more than anything makes its technology and uses dangerous.
As for the, “I, for one, welcome our new computer overlords.” That was what Ken Jennings wrote underneath his Final Jeopardy answer. It’s why he is so wonderfully talented but yet so human. He borrowed the idiom from the Simpson’s but tweaked it to apply to Watson’s performance.
