The Powder Toy

TPT evolution? by perogiepro

perogiepro
perogiepro
99 / 13
24th Aug 2015
22nd Dec 2015
We are the SFPI, we are searching for mechanical life that can detect danger and think. essentialy basic AI. Our end goal is to make one that can replicate and change, so it will """evolve"""
machine intelligence experiment neuron life survival challange evolution spfi

Comments

  • perogiepro
    perogiepro     25th Aug 2015
    Thanks for your suggestions. but im also not so good at this stuff, i just gave the idea. SO ANYONE OUT THERE WE NEED YOU.
  • Technomancer
    Technomancer     25th Aug 2015
    Also, agustrusher is correct. ALL of the recently published 'neural networks' are too random to be much use in generating intelligence, and also do not resemble actual neural networks in their functioning. An AI brain doesn't necessarily have to work the same way as an organic one, but it needs to be largely deterministic rather than random. Even amoebae react to their environments, rather than just randomly jiggling.
  • agustrusher
    agustrusher     25th Aug 2015
    Con.2-HOWEVER, I see where this could be improved, and I hope somebody will use this information, because I'm not very proficient at electronics. What needs to be done, is have connections added from each I/O, that would allow an entity's currect spark status the ability to CHANGE is neighbors spark conductivity (SWCH), NOT it's spark status (though, those connectiosn should remain). In essence, each entity will become a sort of transistor. But of course, this doesn't mean it will magically work
  • Technomancer
    Technomancer     25th Aug 2015
    I have something that might (maybe) be useful here. ID:1844573. It's a CRAY-life brain with an output-limiting device that only allows it to send one signal at a time (multiple outputs will block each other). Also, it's pseudorandom rather than truly random, so you can program it by changing the arrangement of PSCN and SPRK at the start (this will make more sense when you actually see it). Feel free to use it, for this or anything else, if you think it will help.
  • agustrusher
    agustrusher     25th Aug 2015
    Con.1 - These systems, as they are, rely entirely on chaos. It LOOKS like somethings happening. It LOOKS like each entity is dynamically interacting with its neighbors, but it isn't. It's random sparking madness, which seems to ignite that quirky human trait of finding patterns where there are none.
  • agustrusher
    agustrusher     25th Aug 2015
    I'm very open-minded and passionate about TPT Science, but I feel this recent zeal about "neural networks" is unfounded. The general template for most of these TPT networks is entirely unlike real ones; the limitations of TPT reduce the current functionality to something unrecognizable even in principle.
  • Quantum-HE
    Quantum-HE     25th Aug 2015
    It looks not good now...But i will pay attention to it until it become good~
  • perogiepro
    perogiepro     25th Aug 2015
    YES exactly. He is doing great and detects water now. just doesnt react. if you can make something put it in a save for me to put in this one
  • RCAProduction
    RCAProduction     25th Aug 2015
    +1 for the idea, not exactly the save. The bot likes to break... But that's not a problem. If we can get this idea fully functioning, it would be amazing! Perhaps the bot needs some hard wired "instincts" so it knows to avoid stuff. Just a thought.
  • perogiepro
    perogiepro     25th Aug 2015
    @smhlolendevice share the ID of your neural net for me to consider.