IBM's Watson Crushes Harvard, MIT in Jeopardy

<p>
[quote]
On Monday, at the Harvard Business School in Boston, Big Blue’s Watson supercomputer took on a team from the prestigious B School and a second from the MIT Sloan School of Management. Harvard’s Jeopardy team held its own and briefly had the lead. The MIT team, not so much. And in the end, Watson took home the trophy, just as it did when playing two human Jeopardy champs on national television earlier this year.

[/quote]
</p>

<p>IBM</a> Sends Watson Supercomputer to Business School | Wired Enterprise | Wired.com</p>

<p>Must be nice to have a supercomputer.</p>

<p>If Ken Jennings couldn't win, I don't think anyone will.</p>

<p>The time I saw Watson on Jeopardy, it did win. But not because it knew answers (or questions, I guess for Jeopardy) the human contestants did not know. It just seems to be able to buzz in faster than the humans.</p>

<p>Still, its a pretty amazing accomplishment.</p>

<p>I believe how fast Watson buzzes in is determined based on how confident he is on the answer. Nevertheless, its amazing how far computer science has come.</p>

<p>LOL IBM didn't exactly "crush" Harvard if Watson was leading by only $2000 for Final Jeopardy and both teams (Harvard and Watson) got the Final Jeopardy correct.</p>

<p>Nice Exaggeration though!</p>

<p>I think they need to change the rules. Watson buzzes in too fast. When Watson played against Ken Jennings and Brad Rutter, it got questions that I'm sure the humans would have gotten if they were able to buzz in faster (for example, who doesn't know what "sin city" is?). I think that you should be able to buzz in as soon as you have the answer.</p>

<p>don't hate the player, hate the game :P</p>

<p>We've actually talked about Watson a bit in our sensation and perception class. Part of the problem with Watson is that he is, in essence, not playing by the same rules as the other contestants. Watson is not capable of taking visual or auditory input and interpreting that to answer a question. Instead, the questions are imported directly to Watson, at which point he looks for the answer and determines how confident he is in said answer.</p>

<p>I mean is that incredibly surprising with advances in computational power and artificial intelligence algorithms? In most jeopardy games, if all the contestants had unlimited time before hitting the buzzer every single one would know the answer. The person that wins is the person that can hit the buzzer faster, and develop their confidence in the answer faster. Is it really a surprise that humans which do not process information through direct electrical input and subsequent electrical output (all moving at near speeds of light), can be slower? They need sound (slower than speed of light) to hit the ear canal, get converted to an electrical signal, get converted to a chemical signal, back to an electrical signal down the spinal cord, back to chemical signal, back to electrical depolarization, and then biochemical contraction of the buzzing muscle.</p>

<p>It's a parlor trick to simply show off the computational artificial intelligence algorithms developed at IBM.</p>

<p>I'd like to see how Watson does on "The Dating Game." Of course I'm not sure if it's technically a bachelor or a bachelorette. (I'm assuming the young folkis on here even know what the Dating Game was).</p>

<p>With regard to speed of buzzing in, if you watched the jeopardy thing carefully, the humans did the best when they buzzed in before knowing the answer and then thankfully still got it right. The issue is that computers at this point will always "think" faster than people. </p>

<p>The other interesting thing from the jeopardy game was how on that category "words on the keyboard" or something like that, Watson did terribly because it couldn't really process homonyms</p>

<p>A few years ago, I read Braniac, the story of Ken Jennings' amazing run on Jeapordy. Humans, and apparently Watson, have to buzz in before they have fully retrieved the answer. That's why you occasionally see a contestant who has buzzed in but doesn't seem to have a clue. Jennings attributed his string of wins to his learning the timing - if you buzz in too soon, there's a short enforced wait before you can buzz in again. New contestants were much more prone to those errors. It's pretty phenomal that, timing aside, Watson can come up with so many correct answers.</p>

<p>The most unfair part is that Watson gets an electrical signal the moment that anyone can buzz in, while the human contestants have to guess or anticipate when that moment happens. If Watson knows the answer before the moment, they probably design a circuit that buzz in the earliest moment and lock out the rest of the contestants. If you watch the show carefully, everyone was trying to buzz in on most of the questions but Watson wins majority of the time. It becomes a contest of who can buzz in first rather than who is more knowledgeable.</p>

<p>I'd still rather be a person at Harvard than a computer that can only play one game show.</p>

<p>Watson may be a supercomputer but I don't think "he" could pass the SAT.</p>

<p>That's actually a good idea ... Have Watson take the SAT, ACT, GRE or LSAT and see what Watson's score would be.</p>

<p>the reason watson requires a supercomputer is because it has such a large database of information that it needs to access in order to come up with the right keywords. It then assesses the keywords to see which one is most likely the answer. This function would render entirely useless on the SAT. Jeopardy, in essence, is still a game that quizzes your knowledge, not your analytical skills, which is something the SAT does. Although Watson would probably nail the vocabularies, that means Watson would probably go completely blank when he encounters a questions like "train A leaves port A at 6:30, all that stupid logic crap" Still a long ways to go from taking the SAT</p>

<hr>

<p>You know what would level the playing field with the buzzer issue? Have some Japanese roboticists design a mechanism that would physically press the buzzer.</p>

<p>
[quote]
You know what would level the playing field with the buzzer issue? Have some Japanese roboticists design a mechanism that would physically press the buzzer.

[/quote]
</p>

<p>Watson is actually required to press a physical buzzer with an actuator to answer the question.</p>

<p>I think Watson can be programmed to do well in math and those logic questions. Writing section would be absolutely a piece of cake for Watson. But passage problems in critical reading would be the most difficult to do well on for Watson. It is a little different than Jeopardy but with some adjustments, these natural language parsing and understanding can be done. So I think if you give this problem to IBM, I bet Watson can easily ace the SAT.</p>