Interesting article didnt know where it fit best so I wanted to share it here.

  • fubo@lemmy.world
    link
    fedilink
    arrow-up
    5
    ·
    1 year ago

    No, consciousness is just what it feels like when a meat brain uses its meat to change its focus of attention; which gives rise to beliefs (some of them even true!) about a meat brain having a self.

    It takes time, because brains are made of meat, and meat is slow.

    It’s leaky, because brains are made of meat, and meat oozes.

    It generates the image of a “self” because brains are in meat bodies and actually do have physical continuity rather than being disconnected instants of computation; a term for “I, me, myself” is a rough model of the existence of brain features like memory, meat features like hormones, and even ape social-behavior features.

    Attention/awareness is leaky and takes time; meat pumps rhythmically; and chemicals stick around.

    And the meat brain can notice its own meaty doings. Just as it builds models of the outside world, it builds models of itself, with thoughts like “I am in the middle of doing an action” or “I am impatient” or “I feel sleepy” or “OW, LEG CRAMPS SUCK!” That is, its attention can range over not only the leg cramp itself, but its own reaction to having a leg cramp, including how the existence of leg cramps fits into its larger model of whether the world is a terrible place.

    It usually comes up with a lot of correct beliefs out of this reflection, like “this is my leg, not your leg” and “I know English” and “Wow, I am distractable this morning, maybe it’s the strong coffee”. But it also comes up with dubious beliefs like “I am an eternal soul”, “I am fully continuous in time”, or “Oh God, what sin did I commit to deserve this leg cramp?”

    (“This is my leg, not yours” is important because there’s nothing anyone can do to your leg that will make my leg cramp go away. The “self/other” distinction is important to consciousness because it has real-world implications; bodies really are physically disconnected from one another, which is why depersonalization can be an unhealthy thing for a consciousness to do too much.)

    There’s no reason to believe ChatGPT or the like are conscious, because they don’t have the properties that consciousness is a model of. They’re not fed information about their own well-being or place in the world. They don’t observe their own processing. They do run largely as disconnected instants of computation. They don’t live in a space where having a sense of “self/other” is effective.

    (Not yet, anyway. There are folks out there trying to build AI systems that do have the feedback loops that might generate something like consciousness. This is probably a bad idea, and may even be an evil one.)