Page 3 of 8 FirstFirst 12345 ... LastLast
Results 21 to 30 of 73

Thread: AI: Superintelligence - paths, dangers and strategies

  1. #21
    Member
    Type
    INFJ
    Join Date
    Oct 2017
    Posts
    666
    Quote Originally Posted by Architect View Post
    . Tay failed because it didn't have a concept of The Golden Rule. How does an intelligence have this concept? I don't know how you could program that in presently, because you need to give the GAI a semantic understanding of "harm". Which gets to the core problem with present day AI, we have syntactic but not semantic understanding. .

  2. #22
    Senior Member Starjots's Avatar
    Type
    INTP
    Join Date
    Dec 2013
    Posts
    2,086
    Quote Originally Posted by Starjots View Post
    To recap lines of inquiry:

    1. Rules in = output
    Quote Originally Posted by Architect View Post
    Taking it as an exercise let's assume that the present deep neural nets could be evolved into a super intelligence with some additional to be discovered technologies. How would we influence it's development? The easiest would be the same as how we teach children; control the information given to the the GAI. That is give it training examples with the outcome you want. This was shown recently with Tay, a Microsoft Twitter learning chatbot which turned racist after a day of being opened up (they shut it down). People saw an opportunity to 'game' the bot with racist/sexist/Xist tweets and it worked brilliantly. Tay failed because it didn't have a concept of The Golden Rule. How does an intelligence have this concept? I don't know how you could program that in presently, because you need to give the GAI a semantic understanding of "harm". Which gets to the core problem with present day AI, we have syntactic but not semantic understanding.

    So we wound around to the nub of the problem, which is how to encode semantic understanding. Once we have that I think we'll have the next step. NLP seems to be the best field to work on this.
    Isn't Natural Language Processing a hot field of inquiry at the moment?

    Now you've evoked a strange future where humans raise little AI children like their own kids, or at least like dogs or cats. Artificial cow birds so to speak. As long as they are no more capable than dogs (with some sort of body) this would be another harmless diversion for humans.

    Of course, this would be/is industrialized since 'raising' could be done in a short time with preloaded appropriate training data. But won't they all be sort of boring without a screwed up childhood? Then again, boring predictable AIs are probably all for the best at least in the beginning.

  3. #23
    Pull the strings! Architect's Avatar
    Type
    INTP
    Join Date
    Feb 2014
    Location
    West Coast
    Posts
    468
    Quote Originally Posted by Starjots View Post
    Isn't Natural Language Processing a hot field of inquiry at the moment?
    The big excitement is visual processing/convolutional nets, NLP has been doing well too and developing but not getting the attention of visual I think. It's probably a harder problem because again we bump into semantics. Vision is easier and our hardware & algorithms can handle the massive data sets, because all a convnet has to do is say "this is a cat" without having any idea of what a cat is.

    Now you've evoked a strange future where humans raise little AI children like their own kids, or at least like dogs or cats. Artificial cow birds so to speak. As long as they are no more capable than dogs (with some sort of body) this would be another harmless diversion for humans.
    Well that's an analogy with how teach bias to a DNN. If you want a biased DNN, feed it a biased training set. Theoretically you could instead go in and twiddle the weights but that's too hard.

    Of course, this would be/is industrialized since 'raising' could be done in a short time with preloaded appropriate training data. But won't they all be sort of boring without a screwed up childhood? Then again, boring predictable AIs are probably all for the best at least in the beginning.
    I don't know, like I said we're missing some important stuff still.

  4. #24
    Pull the strings! Architect's Avatar
    Type
    INTP
    Join Date
    Feb 2014
    Location
    West Coast
    Posts
    468
    Quote Originally Posted by mara View Post
    ...skynet...
    People make that analogy all the time but I think its too easy. Assuming AI necessarily turns evil, Skynet took over because they hooked it up to 'everything', which is already an impossibility in the present day world. Second, in the present world there wouldn't be one Skynet, there would be many with varying degrees of alignment (good-evil) and scope (=control). Not to mention all the guys like me who will have their own Skynets operating out of their garages (yes I build AI's in my garage for phun and profit)

  5. #25
    Member
    Type
    INFJ
    Join Date
    Oct 2017
    Posts
    666
    Quote Originally Posted by Architect View Post
    People make that analogy all the time but I think its too easy. Assuming AI necessarily turns evil, Skynet took over because they hooked it up to 'everything', which is already an impossibility in the present day world. Second, in the present world there wouldn't be one Skynet, there would be many with varying degrees of alignment (good-evil) and scope (=control). Not to mention all the guys like me who will have their own Skynets operating out of their garages (yes I build AI's in my garage for phun and profit)
    there is already a 'skynet'. it is a military satellite network

    As for things being hooked up to everything that is what is now occurring with the creation of the 'SMART grid' and 'internet of things'. Also wifi is being pushed everywhere now so that all the interconnected devices can communicate with each other.

    Who is to say that AI isn't already connected to the internet and learning from us?

    inter-NET.....sky-NET

  6. #26
    Pull the strings! Architect's Avatar
    Type
    INTP
    Join Date
    Feb 2014
    Location
    West Coast
    Posts
    468
    Quote Originally Posted by mara View Post
    As for things being hooked up to everything that is what is now occurring with the creation of the 'SMART grid' and 'internet of things'.
    Connection doesn't equal control. Your brain is connected (and has control) over the rest of your body, but you can't mentally turn off your period.

    If you're interested in these matters prominent researchers (and people with an opinion like Elon Musk) gathered this year at Asilomar to create principles of beneficial AI, here's one of the panels



    The principles are here

    I believe @Starjots intended the discussion about the how of beneficial AI, but maybe we could discuss if AI is necessarily good or evil, or if we can have control at all.
    Last edited by Architect; 11-22-2017 at 02:48 PM. Reason: accidentally hit send

  7. #27
    Member
    Type
    INFJ
    Join Date
    Oct 2017
    Posts
    666
    Quote Originally Posted by Architect View Post
    Connection doesn't equal control. Your brain is connected (and has control) over the rest of your body, but you can't mentally turn off your period.

    If you're interested in these matters prominent researchers (and people with an opinion like Elon Musk) gathered this year at Asilomar to create principles of beneficial AI, here's one of the panels

    The principles are here

    I believe @Starjots intended the discussion about the how of beneficial AI, but maybe we could discuss if AI is necessarily good or evil, or if we can have control at all.
    Elon Musk has actually warned against the dangers of AI

    His proposed solution to this is that we all have neural laces implanted into our brains so that our minds can become one with the AI

    ie the borg

  8. #28
    Pull the strings! Architect's Avatar
    Type
    INTP
    Join Date
    Feb 2014
    Location
    West Coast
    Posts
    468
    Quote Originally Posted by mara View Post
    Elon Musk has actually warned against the dangers of AI
    That's what I said, he has an opinion which he has been very public about.

    His proposed solution to this is that we all have neural laces implanted into our brains so that our minds can become one with the AI
    It's a good idea, the thought is that if we can operate at AI speeds we can be a counterbalance.

    Speaking of which, I have half an idea that the first GAI's will actually be us, via this connectivity. We are already cyborgs and are the only creature to actively augment ourselves with technology. And from a certain perspective the internet is already alive and intelligent, just via the distributed and competing human intelligences.

    Edit: the only thing interesting about Mr Musk (to me) is his probable type - ISTP. He read a lot of scifi as a kid and his approach in his work is all in the physical - transportation mainly. He's distrustful of the non physical like AI. His first wife is probably an NT by the way, or at least appears to be so to me - ?NTJ I believe.

  9. #29
    Member
    Type
    INFJ
    Join Date
    Oct 2017
    Posts
    666
    Quote Originally Posted by Architect View Post
    That's what I said, he has an opinion which he has been very public about.



    It's a good idea, the thought is that if we can operate at AI speeds we can be a counterbalance.

    Speaking of which, I have half an idea that the first GAI's will actually be us, via this connectivity. We are already cyborgs and are the only creature to actively augment ourselves with technology. And from a certain perspective the internet is already alive and intelligent, just via the distributed and competing human intelligences.
    except you won't keep pace with AI

    It will simply control your brain

    you will be assimilated into the borg

  10. #30
    know nothing pensive_pilgrim's Avatar
    Type
    INTP
    Join Date
    Dec 2013
    Posts
    4,179
    Quote Originally Posted by mara View Post
    except you won't keep pace with AI

    It will simply control your brain

    you will be assimilated into the borg
    How could an AI want anything other than what a human has programmed it to want?

Similar Threads

  1. Tea strategies
    By Blorg in forum The Pub
    Replies: 46
    Last Post: 05-05-2017, 02:54 PM
  2. People's Rubic's Cube strategies
    By BarIII in forum The Pub
    Replies: 20
    Last Post: 08-09-2016, 04:28 PM
  3. Foolishly reveal your password creation strategies
    By notdavidlynch in forum The Pub
    Replies: 41
    Last Post: 09-15-2014, 09:34 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •