Stories
Slash Boxes
Comments

SoylentNews is people

posted by Fnord666 on Saturday February 03 2018, @09:48AM   Printer-friendly
from the modern-day-punishment dept.

Fire good. AI better:

Google CEO Sundar Pichai says artificial intelligence is going to have a bigger impact on the world than some of the most ubiquitous innovations in history. "AI is one of the most important things humanity is working on. It is more profound than, I dunno, electricity or fire," says Pichai, speaking at a town hall event in San Francisco in January.

A number of very notable tech leaders have made bold statements about the potential of artificial intelligence. Tesla boss Elon Musks says AI is more dangerous than North Korea. Famous physicist Stephen Hawking says AI could be the "worst event in the history of our civilization." And Y Combinator President Sam Altman likens AI to nuclear fission.

Even in such company, Pichai's comment seems remarkable. Interviewer and Recode executive editor Kara Swisher stopped Pichai when he made the comment. "Fire? Fire is pretty good," she retorts. Pichai sticks by his assertion. "Well, it kills people, too," Pichai says of fire. "We have learned to harness fire for the benefits of humanity but we had to overcome its downsides too. So my point is, AI is really important, but we have to be concerned about it."

Also at CNN and Gizmodo.


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 5, Insightful) by bradley13 on Saturday February 03 2018, @10:31AM (23 children)

    by bradley13 (3053) on Saturday February 03 2018, @10:31AM (#632457) Homepage Journal

    I've worked in the field of AI off and on since, gawd, 1985. As part of my studies in 1985, we looked at the history of AI, which actually goes all the way back to the 1950s. Perceptrons were first discussed in 1957. Rule-based systems weren't formalized until the 1970s, but only because people took predicate logic for granted, and wrote rule-based systems without putting labels on them.

    Sort of like fusion, the AI breakthrough is always just a few years away. And yet, somehow we are no closer now than we have ever been. The only reason that AI is more successful today than it was in the 1950s is the sheer amount of computing power that we can now throw at it. Somewhere, we are still missing a fundamental clue.

    --
    Everyone is somebody else's weirdo.
    Starting Score:    1  point
    Moderation   +3  
       Insightful=2, Interesting=1, Total=3
    Extra 'Insightful' Modifier   0  
    Karma-Bonus Modifier   +1  

    Total Score:   5  
  • (Score: 2, Disagree) by maxwell demon on Saturday February 03 2018, @10:51AM (12 children)

    by maxwell demon (1608) on Saturday February 03 2018, @10:51AM (#632469) Journal

    The development of AI seems to go along this path:

    1. Some problem is identified that needs to be solved if one wants AI.
    2. After some time, that problem indeed gets solved.
    3. As soon as the solution is found, it is quickly asserted that the solution is no true AI.

    Of course if you continuously move the goalpost as soon as you approach it, you'll never reach it.

    --
    The Tao of math: The numbers you can count are not the real numbers.
    • (Score: 5, Insightful) by c0lo on Saturday February 03 2018, @11:10AM (3 children)

      by c0lo (156) Subscriber Badge on Saturday February 03 2018, @11:10AM (#632477) Journal

      I'm sorry, no, Strong AI [wikipedia.org] is defined for quite some time.
      And we aren't much closer to it.
      Bu I can see why CEO with a finger in massive NNs today are very happy to claim AI - works wonders for the stock value.

      --
      https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
      • (Score: 2) by Gaaark on Saturday February 03 2018, @01:19PM (1 child)

        by Gaaark (41) on Saturday February 03 2018, @01:19PM (#632510) Journal

        I prefer to call it Strong Blockchain: seems to do better for IPO's.
        ;)

        --
        --- Please remind me if I haven't been civil to you: I'm channeling MDC. ---Gaaark 2.0 ---
        • (Score: 2) by maxwell demon on Saturday February 03 2018, @05:12PM

          by maxwell demon (1608) on Saturday February 03 2018, @05:12PM (#632597) Journal

          Artificial Blockchain Intelligence? :-)

          --
          The Tao of math: The numbers you can count are not the real numbers.
      • (Score: 2) by maxwell demon on Saturday February 03 2018, @05:19PM

        by maxwell demon (1608) on Saturday February 03 2018, @05:19PM (#632603) Journal

        Yes, but nobody (except you, apparently) was talking about strong AI. Saying only strong AI is true AI is like saying all those industrial robots are no robots, because they don't look anything like the robots in SF movies.

        --
        The Tao of math: The numbers you can count are not the real numbers.
    • (Score: 2) by Runaway1956 on Saturday February 03 2018, @11:16AM (7 children)

      by Runaway1956 (2926) Subscriber Badge on Saturday February 03 2018, @11:16AM (#632478) Journal

      Artificial intelligence seems at least moderately well defined. We built it, and it's capable of assessing a situation, and coming to a decision on a course of action, on it's own, without any input from humans being necessary. An AI can and will disagree with it's creators sometimes, and make arguments in favor of it's chosen course of action. An AI will no longer be dependent on humans to make decisions. Artificial intelligence.

      That AI may be dependent on us for energy, or for supplies, or whatever. It may need us for "feel good" - that is, to satisfy some need for a sense of a accomplishment.

      Most things that we call AI today are pretty feeble attempts, in that light. A complex program is still just a program, designed to perform a set of tasks, then to shut down and await further input. Even lower animals are more "intelligent", in that, they can continue all on their own, without any intervention from mankind. Eat, sleep, defecate, make home more comfortable, eat, sleep, defecate, defend from predator, copulate, rinse and repeat endlessly.

      Maybe the AI people need to introduce the computers to sex to get things kicked off?

      • (Score: 2) by maxwell demon on Saturday February 03 2018, @12:58PM (3 children)

        by maxwell demon (1608) on Saturday February 03 2018, @12:58PM (#632502) Journal

        Humans don't function well on their own. Total isolation is considered torture for a good reason.

        --
        The Tao of math: The numbers you can count are not the real numbers.
        • (Score: 3, Funny) by Gaaark on Saturday February 03 2018, @01:22PM (2 children)

          by Gaaark (41) on Saturday February 03 2018, @01:22PM (#632513) Journal

          Shit, give me a good computer and the internet and I'm fine!

          Ooooo, yeah. Sex.

          Never mind....

          --
          --- Please remind me if I haven't been civil to you: I'm channeling MDC. ---Gaaark 2.0 ---
          • (Score: 0) by Anonymous Coward on Saturday February 03 2018, @03:18PM (1 child)

            by Anonymous Coward on Saturday February 03 2018, @03:18PM (#632546)

            no, the computer can come with interfaces for that sort of network penetration as well.

            • (Score: 2) by Gaaark on Saturday February 03 2018, @06:40PM

              by Gaaark (41) on Saturday February 03 2018, @06:40PM (#632627) Journal

              Tell me more about this 'penetration', big momma........

              --
              --- Please remind me if I haven't been civil to you: I'm channeling MDC. ---Gaaark 2.0 ---
      • (Score: 2) by chromas on Saturday February 03 2018, @06:53PM

        by chromas (34) Subscriber Badge on Saturday February 03 2018, @06:53PM (#632631) Journal

        It sounds like the problem is we haven't created a pooping AI yet.

      • (Score: 2) by Grishnakh on Sunday February 04 2018, @04:55PM (1 child)

        by Grishnakh (2831) on Sunday February 04 2018, @04:55PM (#632954)

        Even lower animals are more "intelligent"

        Lots of animals (dogs, cats, etc.) are not only quite intelligent, but have definite personalities, and do lots of things besides the necessities of survival, which don't really benefit their survival, and just because they want to and it amuses them, just like we do.

        • (Score: 2) by Runaway1956 on Sunday February 04 2018, @06:28PM

          by Runaway1956 (2926) Subscriber Badge on Sunday February 04 2018, @06:28PM (#632976) Journal

          Well, yeah - but we haven't nearly reached that level with AI yet. I was thinking more about lower animals. Paramecium and such, except most of those reproduce asexually, so they have no need of copulation. Hmmmmm - do asexual creatures indulge in sexual conduct? Poking a little fun, knowing that they'll never be taken seriously?

  • (Score: 2) by c0lo on Saturday February 03 2018, @11:02AM

    by c0lo (156) Subscriber Badge on Saturday February 03 2018, @11:02AM (#632473) Journal

    The only reason that AI is more successful today than it was in the 1950s is the sheer amount of computing power that we can now throw at it. Somewhere, we are still missing a fundamental clue.

    What we are missing is a true AI.
    For the time being, all we have are sophisticated classifiers - not even particularly robust ones [blog.xix.ai]

    The para I like the best in the linked:

    Real-world examples

    It’s a long way from our experiments to DEFCON keynote speech. Yet even now the range of possibilities for adversarial attacks is worrisome. Just to name a few ones inspired by the paper “Adversarial examples in the physical world” (strictly for discussion purposes):

    1. Print a “noisy” ATM check written for $100 — and cash it for $1000000.
    2. Swap a road sign with a slightly perturbed one that would set the speed limit to 200 — in a world of self-driving cars it can be quite dangerous.
    3. Don’t wait for self-driving cars — redraw your license plate and cameras will never recognize your car.
    --
    https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
  • (Score: 0) by Anonymous Coward on Saturday February 03 2018, @11:05AM (6 children)

    by Anonymous Coward on Saturday February 03 2018, @11:05AM (#632475)

    The only reason that AI is more successful today than it was in the 1950s is the sheer amount of computing power that we can now throw at it. Somewhere, we are still missing a fundamental clue.

    Maybe thats your clue? Along with training the AI for 10-20 years, I dont see why that couldnt be the solution.

    • (Score: 3, Interesting) by Arik on Saturday February 03 2018, @11:51AM (5 children)

      by Arik (4543) on Saturday February 03 2018, @11:51AM (#632488) Journal
      "Maybe thats your clue? Along with training the AI for 10-20 years, I dont see why that couldnt be the solution."

      "Couldn't be the solution?" There's so much packed into that phrasing, and so inappropriate.

      Can we absolutely rule out, a priori, the notion that simply throwing resources at tasks we do not understand will magically result in AI? No more than you can absolutely rule out the notion that the universe was created in a flash by the will of the Flying Spaghetti Monster.

      It's ludicrously bad thinking. Computers are big calculators. Intelligence is not artificial, it's a quality of (some) programmers, not of computers. When you build a ludicrously complex system you don't understand, feed it a bunch of numbers and then obey its output, that's not artificial intelligence, it's just modern superstition.
      --
      If laughter is the best medicine, who are the best doctors?
      • (Score: 2) by fyngyrz on Saturday February 03 2018, @11:59AM (4 children)

        by fyngyrz (6567) on Saturday February 03 2018, @11:59AM (#632490) Journal

        When you build a ludicrously complex system you don't understand, feed it a bunch of numbers and then obey its output, that's not artificial intelligence, it's just modern superstition.

        So, you're talking about children, then.

        • (Score: 0) by Anonymous Coward on Saturday February 03 2018, @01:37PM

          by Anonymous Coward on Saturday February 03 2018, @01:37PM (#632515)

          > ... and then obey its output, ...

          Well, my mother (88) certainly fed me a bunch of numbers, letters, and a lot of good food. But she's not ready to obey my output yet (except in a few specialized domains like figuring the tip at a restaurant).

        • (Score: 2) by Arik on Sunday February 04 2018, @02:28AM (2 children)

          by Arik (4543) on Sunday February 04 2018, @02:28AM (#632752) Journal
          Children may be complex systems that we did not create, but they are not reducible to a calculator. In fact, they don't calculate at all. Sure, most of them can be taught to do it, but they're very slow and error prone even then.

          That's why we invented computers in the first place. As long as you understand what you're asking them, they can give very good answers, very quickly.

          But when you don't know what you're asking then the answer is, for all intents and purposes, gibberish as well.

          GIGO.

          --
          If laughter is the best medicine, who are the best doctors?
          • (Score: 2) by fyngyrz on Sunday February 04 2018, @12:13PM (1 child)

            by fyngyrz (6567) on Sunday February 04 2018, @12:13PM (#632885) Journal

            Children may be complex systems that we did not create, but they are not reducible to a calculator. In fact, they don't calculate at all. Sure, most of them can be taught to do it, but they're very slow and error prone even then.

            No, no. I didn't specify young children.

            To belabor the point, the point is: we don't understand, and cannot predict, humans. And while you are right that humans are not reducible to calculators, that only drives the point home even further: You cannot predict what a human will do. Humans are indeed ludicrously complex active systems we have very little control over that we do indeed obey and depend upon the output of; we apply reason to that to some degree (well, some of us do) but the same can be said of LDNLS [fyngyrz.com], and no doubt, AI when (or if) it ever gets here.

            • (Score: 2) by Arik on Sunday February 04 2018, @07:33PM

              by Arik (4543) on Sunday February 04 2018, @07:33PM (#632998) Journal
              Humans are unpredictable.

              The oracle of rat-bones is similarly unpredictable.

              Therefore the rat-bone oracle is AI?
              --
              If laughter is the best medicine, who are the best doctors?
  • (Score: 0) by Anonymous Coward on Saturday February 03 2018, @02:15PM

    by Anonymous Coward on Saturday February 03 2018, @02:15PM (#632524)

    The only reason that AI is more successful today than it was in the 1950s is the sheer amount of computing power that we can now throw at it.

    Yeah all that human computing power via those captchas.

    Somewhere, we are still missing a fundamental clue.

    And the way things are being done nowadays even if they succeed in making a major AI breakthrough they might not even understand it. Keep dumping random stuff into a cauldron and one day somehow it works and nobody knows why :).

    The current state of AI is like where Alchemy was before Chemistry and the other sciences.

  • (Score: 1, Insightful) by Anonymous Coward on Saturday February 03 2018, @04:23PM

    by Anonymous Coward on Saturday February 03 2018, @04:23PM (#632568)

    So you entered the field just when the irrational exuberance of that time had worn off, and mentioning AI would get you a laugh and a kick in the ass out the door.
    I predict we'll have another bust, and we'll go back and call this work machine learning and pattern recognition again, because the term AI is poison for funding.