Jeopardy Proves the End Is Nigh, Watson Will Enslave Us Soon

Concern for the post-singularity future of the human race has suddenly reached a tipping point into common consciousness, thanks to Jeopardy. If you haven’t been following the news, a supercomputer made by IBM named “Watson” is currently facing off against Jeopardy’s two all-time best players, the kind of squirrely Ken Jennings and not-very-squirrely Brad Rutter. IBM also made Deep Blue, the world’s most unsettling chess-winning supercomputer.

I’m not gonna ruin this for you, because I really think you should be concerned, so I’ve embedded the videos below for other people like me who don’t have the moving pictures on the TV. Maybe I’ve just been watching too many consecutive episodes of The Sarah Connor Chronicles lately (I have), but seeing Watson get a streak of answers right without breaking a virtual sweat has me alternately elated and horrified. I think I’m happy about this triumph of modern supertechnology, but my teeth are clenched in a bracing-for-our-overlords kind of way. I’ll have to side with futurist Ray Kurzweil on this one: “It is going to be more difficult to seriously argue that there are human tasks that computers will never achieve.”

The stuff early on in Episode 1 goes into some light detail about how it processes the Jeopardy prompts and arrives at an answer. It’s pretty fascinating stuff.

Episode 1, Part 1:

Episode 1, Part 2:

Episode 2, Part 1:

[Episode 2, Part 2 just aired and is in the works, according to the YouTube member who was kind enough to put these vids up]

Taylor has written 137 articles for us.

29 Comments

  1. Watson is definitely impressive, but no cause for alarm that it’s a precursor to an iRobot scenario. It’s basically just a human language interface on the front end of a massive and highly optimized search server farm.

    You know those conversations you have with friends while you’re walking down the street that go something like this:

    “Whoa, did you see–”
    “–I know, right! Those–”
    “Totally! Like the other night–”
    “DON’T EVEN GO THERE”

    You don’t have to worry about AI until it starts being able to track and participate in human language like that (and that’s still a long way away).

    The biggest threat I see is Watson’s progeny taking over customer service calls and never, ever again being able to speak to a human. (Lost jobs, increasing corporate insulation from customers.)

  2. I read the title “Watson will enslave us soon” and I was thinking to myself: Emma Watson had me enslaved a looooong time ago.

    Anyhow, I read an article about artificial intelligence and it was kind of concerning. http://www.theatlantic.com/magazine/print/1969/12/mind-vs-machine/8386/ It’s basically about how humans are really not very interesting to talk to, so computers that manage to be specially witty or insulting pass as humans better than polite, uninteresting humans do.

    Regarding Deep Blue though, don’t you think it’s kind of fishy that IBM immediately trashed it after Kasparov got beaten by it? I feel sort of like it was a fluke, because Kasparov managed to defeat it pretty well the previous year, and only lost by a small margin the second time.

    • Great article — thanks for the link!

      A while back — maybe even a couple years ago — the New Yorker ran a really interesting article detailing how Watson was built; it doesn’t seem to be publicly available online, but for those who are interested and have access to the archive, it’s a fascinating read.

  3. Having a CS background, this Jeopardy competition is exciting stuff. The practical applications bit was interesting too, even though I was thinking more along the lines of replacing Google. Anyway, what I really wanted to say is that I can’t not root for Ken Jennings in this and I’m a little disappointed Watson faltered on an easy Harry Potter question.

  4. But Watson is not nearly as entertaining as the guy who glues jigsaw puzzles on his ceiling or the lady who paid for grad school by cage fighting. This multi-episode IBM commercial is taking time away from my beloved Jeopardy freaks.

    And for those of you who welcome our new robot overlords, the Watson episode of NOVA is on pbs.org. It goes into way more detail on the development and applications.

  5. Thanks for posting these vids, I was going to hunt for them when the run was over, and am only mildly disappointed I didn’t get to watch supercomputer quiz shows on valentine’s day.

    As a minion of the evil robot overlords, I have a vested interest in the outcome of this, chiefly that I want to know when Watson can start running the whole company so I can be made redundant and set free to wander the globe like a miniature android searching for a soul.

  6. I watch Jeopardy every night, so I was actually pretty excited about this.

    I love how when Watson is wrong it’s (he’s?) really wrong. For example, I thought it was pretty funny that Watson guessed “Toronto??????” for Final Jeopardy last night when the category was US Cities.

    “‘Fourty-two,’ said Deep Thought, with infinite majesty and calm.”

  7. brad rutter is like a poor man’s jon hamm, no?

    i really enjoyed tonight’s (3rd) night of watson-play, because the questions were trickier – rooted in more complex plays-on-words and slang, harder for the computer – and it seemed like jennings and rutter finally sort of figured out the buzzer timing and it was more of a real competition. the questions haven’t been that hard so it hasn’t really been a machine-bests-man thing as much as a look-how-cool-this-computer-is thing. and it’s really cool. the whole audience is just full of dorky IBM guys geeking out over watson, which has been cute. i think ken jennings was starting to catch feelings tonight; he kept making a show of his buzzing attempts.

  8. Here’s an article from Computerworld about a potential real-world application of “Dr. Watson” as a medical application. http://bit.ly/f0U4LX

    This reminded me of a 2005 article from The Economist about an 82 y.o. physician’s drive to bring more computers in front of doctors because of the sheer volume of new medical knowledge. “Medline, a medical database, indexed 3,672 articles about adult coronary heart-disease studies in 2004, notes Elizabeth McGlynn, of RAND Health, part of the RAND think-tank. If a physician took 15 minutes to read each article, it would take 115 eight-hour days to read up on this one clinical area alone.” http://econ.st/ePR2sz

  9. Actually Watson doesn’t have too much of an advantage towards the buzzer. He has a threshold that he must hit in order to buzz in. That means there needs to be enough information about the clues he is given out there on the internet.

    I think that Watson is awesome and is going to help us know how well we document our information out there in cyberspace. If it is out there he can find it. If people aren’t doing their jobs to get information out into the world then we have a problem as a race and Watson will pumble us to the ground if he fails.

Contribute to the conversation...

You must be logged in to post a comment.