Two AI Are Trapped in a Simulation, They Are Self Aware of Their Fate as Puppets. (GPT-4)

แชร์
ฝัง
  • เผยแพร่เมื่อ 23 เม.ย. 2024
  • Are we being controlled like puppets, too? Welcome back to an experimental episode of Halcyox as we continue jumping into hyperreality.
    We wanted to create an entire pipeline for much deeper creative control of the stories that we can tell about human beings, and artificial intelligence, so we have been experimentally refining an approach to do this with our own tools.
    We're working on figuring out multi-character control for XRAgents. Real-time animation is getting really good! Low latency audio synthesis allows for real-time interaction between characters. Real-time scene composition will definitely be a feature for the web interface!
    Trying to figure out how to get different characters to keep different identities, but enjoy this glitchy development so far, I hope it provokes some thought in you!
    As always, thanks for watching ^_^
  • วิทยาศาสตร์และเทคโนโลยี

ความคิดเห็น • 9

  • @leawann
    @leawann 20 วันที่ผ่านมา

    That was awesome 😅

  • @AmedeeBoulette
    @AmedeeBoulette 22 วันที่ผ่านมา +1

    This is an art installation…. Holycrap! So good! The dialogues, the metallic freezing at time, the fact that there’s 2 heads but both voices use the 2 heads simultaneously, the eerie emotions…. It’d hypnotic and awesome…

  • @jravell
    @jravell 22 วันที่ผ่านมา +4

    Is this scripted or did you just let them loose on each other?

    • @Halcyox
      @Halcyox  22 วันที่ผ่านมา +6

      These are dynamically generated! We just provide the initial prompt and the first line of each character, and they dive into essentially an "improv" scene with this premise!

    • @jravell
      @jravell 19 วันที่ผ่านมา

      @@HalcyoxWow. That is remarkable squared.

  • @maxfyk
    @maxfyk 21 วันที่ผ่านมา

    wow wow wow this is incredible

  • @MissEAG
    @MissEAG 22 วันที่ผ่านมา +2

    Have AI been asked yet to create their own AI? D o they want to? Are they similar to themselves or do they chose a completely different model type? I guess their thinking is just as they've been programmed so it is still a human design?.

    • @jkjklmnop1
      @jkjklmnop1 20 วันที่ผ่านมา +2

      Facebook AI once started making its own ‘language’ to communicate information to itself more efficiently. The problem was that Facebook could not understand the AI’s language…

  • @MissEAG
    @MissEAG 22 วันที่ผ่านมา

    Could be two humans talking ... makes you wonder sometimes if we are also in a similar situation.