One brain in a vat... no..make it two

Discussion in 'Computer Science & Culture' started by Quantum Quack, May 27, 2023.

  1. Quantum Quack Life's a tease... Valued Senior Member

    Messages:
    23,328
    Was wondering what would be the long term outcome of placing two, independent of each other, AI's in a chamber with learning capacity. Set them off and come back in say 12 months to assess results....
    Example:
    Droid A teaching droid B which is teaching droid A with both droids teaching themselves from the out put of the other droid. ( Give each droid a read only access to the WWW)
     
  2. Google AdSense Guest Advertisement



    to hide all adverts.
  3. cluelusshusbund + Public Dilemma + Valued Senior Member

    Messages:
    7,985
    When the branes are "set off" ... an they are identical... wit identical envrioments... after A teeches B what it knows... an then B teeches A what it knows... etc... etc... they woud each take turns at havin more knowledge than the other.!!!

    Next

    Please Register or Log in to view the hidden image!

     
    Quantum Quack likes this.
  4. Google AdSense Guest Advertisement



    to hide all adverts.
  5. Quantum Quack Life's a tease... Valued Senior Member

    Messages:
    23,328
    say...one was specialist in physics and the other mathematics

    a TOE in the making perhaps

    Please Register or Log in to view the hidden image!

     
  6. Google AdSense Guest Advertisement



    to hide all adverts.
  7. cluelusshusbund + Public Dilemma + Valued Senior Member

    Messages:
    7,985
    Why not... wit AI some very interestin times may be closer than we thank

    Please Register or Log in to view the hidden image!

     
    Quantum Quack likes this.
  8. C C Consular Corps - "the backbone of diplomacy" Valued Senior Member

    Messages:
    3,324
    Two or more antagonistic learning systems reciprocally regulating each other (competing) may be what will parsimoniously allow a solipsistic version of simulated reality to be possible in the future. Where coherent environments can be created on the fly for observers. Something that the brain already does in dreaming, but can be sloppy at when it comes to internal consistency. (I.e., the brain could use an adversarial critic.)

    Technological solipsism has to be more lawful and coherent than dreams. More so than (what might be construed as) an early ancestor in the video below. But it's a matter of incremental improvement over the coming decades.

    Old-fashioned biological dreams are still pretty impressive, though, from the standpoint of the brain generating and maintaining them improvisationally from formula (vaguely echoing the forms of Plato?). Rather than having to literally store and maintain a "whole" simulated world specifically for that "story", even when most of it would not be interacting with the roving POV (observer).

    Generative adversarial network (GAN)
    https://en.wikipedia.org/wiki/Generative_adversarial_network

    GAN Theft Auto ..... LINK
     
    Last edited: May 27, 2023
    Quantum Quack likes this.
  9. Baldeee Valued Senior Member

    Messages:
    2,226
    I would suggest you watch the classic film Colossus: The Forbin Project which gives a 70's answer to just this question, albeit with a militaristic slant.
    Great film.
     
    Quantum Quack likes this.
  10. Baldeee Valued Senior Member

    Messages:
    2,226
    Here's a thought: a VR head-set that reads your thoughts and procedurally creates a virtual world in response.
    It's AI, so you can train it through likes / dislikes of what you're seeing.
    There may be a selection of preset world-types, so abstract, steampunk, medieval, contemporary urban etc, which really just give it some ground rules and starting ideas.
    It doesn't create what you're thinking of, but somehow creates in response to what you're thinking.

    Not sure if that would even work as an idea, though?
     
    C C likes this.
  11. mathman Valued Senior Member

    Messages:
    2,002
    I suggest a minimal approach. Both AI have the union of their ability - no more.
     
    Quantum Quack likes this.
  12. Write4U Valued Senior Member

    Messages:
    20,069
    I am under the impression that all GPT AI apps are controlled by only a few master servers, located in separate buildings.
    So I assume that technically they all could be connected and become a single emergent master server (1 brain in a vat).

    Please Register or Log in to view the hidden image!

    Please Register or Log in to view the hidden image!


    Microsoft Improves Position In Artificial Intelligence With GPT-4 --- GPT-3 and beyond - Part 1: The basic recipe

    https://platform.openai.com/docs/guides/production-best-practices
    https://dida.do/blog/gpt-3-and-beyond-part-1
     

Share This Page