• sj_zero
    link
    fedilink
    21 year ago

    Feels like there’d be ethical implications to growing brain tissue to work in our disposable electronics.

    • SalamanderOPM
      link
      fedilink
      21 year ago

      Yes, for sure. The field of AI is also tasked with solving related ethical issues. To figure out what those implications are we still need to learn more about thinking.

      But I think that this work is not really so much about the practical use of integrating a neuron into an electronic device, but rather is a paper about how to train a neuron.

      It is actually a very deep and interesting concept. If we think about how one can train a dog via positive and negative reinforcement, that concept is obvious - the dog will change its behavior to try to minimize pain and maximize reward. But the dog is already possesses a complex system to interpret “pain” and “reward”.

      As I understand it, this work is about taking a neural system in its simplest form and answering the question “what is ‘positive’ and ‘negative’ reinforcement for a neuron?” The neuron is just a neuron, so it doesn’t have a concept of pain in the same sense that an animal does. But there is a theory that states that neural systems do experience a form of “pain” and that this “pain” is caused by unpredicatable feedback, so the neural network will behave in ways that minimize the amount of unpredicatable feedback. It is a fascinating concept, and the evidence for this model that this ping-pong experiment provides is very compelling.

      • sj_zero
        link
        fedilink
        21 year ago

        I agree with you on all this, I’m more concerned because the abstract of the article is so breathlessly excited about the practical applications of the research.

        As basic research it’s important by itself.