Seems like researchers are getting uncomfortably good at mimicking consciousness- so much so that it’s beginning to make me question the deepest parts of my own brain. Perhaps the difference between AI and myself is that I am only prompted by outward stimuli? It seems as though that is what makes the intelligence “artificial”

Christ even as I type this, my thought processes mimic, for example, deepseek’s exposed deep think capabilities. Fuck idk if I’ll be able to unsee it. Seems like the only thing we have on it right now is emotion and even that seems to be in danger

My final takeaway from all of this is that we are in hell

  • NaevaTheRat [she/her]@vegantheoryclub.org
    link
    fedilink
    English
    arrow-up
    50
    ·
    4 days ago

    Lots of things can look similar from some perspective while being nothing alike. All of stage magic rests on this. Conversely many things which look very different can serve the same purpose. Human eyes and a digital camera will both form images through extremely different mechanisms.

    We don’t know how it is that consciousness arises. It could be extremely widespread and you could be the only conscious being in all of time. You cannot prove that I am conscious and I cannot prove that you are.

    That said, most people think brains are involved given that messing with them (either chemically or physically) causes people to report changes in consciousness. Of course we just have to take at their word that the humans reporting this are conscious ;). Brains, even non human brains, to a staggering amount of stuff and are amazingly complex; there are around 100 billion neurons and 100 trillion connections in a human brain and despite sharing names NN ‘neurons’ are much simpler machines than wetware ones.

    It could be that statistical inference of next token in language is the or at least one of the systems that causes consciousness to arise but we would just be guessing. Many people tend to privilege behaviours we feel other animals don’t share as indicators of consciousness for extremely shakey reasons and so a machine that emits stuff which looks like human language is spooky.

    tldr: worry about the other stuff on this earth that has way more in common with how you work than a computer before you worry about a computer. But maybe worry about computers too, nobody knows. Still if you think a computer is more likely to be conscious at this stage than say a mouse you are probably deeply confused.

    • dat_math [they/them]@hexbear.net
      link
      fedilink
      English
      arrow-up
      27
      ·
      4 days ago

      Still if you think a computer is more likely to be conscious at this stage than say a mouse you are probably deeply confused

      And even then, if you can be convinced that consciousness can arise/emerge in an llm, you really ought to give the mouse (and every animal with a nervous system) the benefit of the doubt.

      • NaevaTheRat [she/her]@vegantheoryclub.org
        link
        fedilink
        English
        arrow-up
        16
        ·
        4 days ago

        Yeah I mean I’m vegan because I realised that I kept moving my line of ‘likely to be conscious’ as more and more complex behaviours became well documented. Then I realised ‘Creatures with nervous systems that work roughly the same as mine are not conscious’ was a really bizarre null hypothesis to have.

        Maybe these machines are but we really need to keep in mind that they are really simple compared to living creatures like us and the way they emit text is nothing like the way we think humans do. They literally just predict the next token (often a word fragment) based on statistical inference from aggregated text. How they do it is very complex, how they are trained is quite interesting, but what they do is far more linear and simple than anything a human brain does.

        Putting aside the ways in which they’ve been made and the economic environment they exist in they’re super cool. It’s fascinating you can get roughly coherent very convincing speech-like emissions this way. It’s thought provoking and tantilises with various research possibilities and challenges what we think might be required to make something that at least appears smart. That’s neat as fuck, and if they weren’t being used to try and destroy the world and overload the power grid hey awesome, good job researchers.

        But they are doing something vastly different from how life works at a much simpler and shallower level. There aren’t good reasons to suspect there is more to them than fancy illusion at this stage.

        • cosecantphi [he/him, they/them]@hexbear.net
          link
          fedilink
          English
          arrow-up
          3
          ·
          edit-2
          4 days ago

          That’s exactly where I am on this! I used to think maybe humans were special, then I thought maybe animals were special, then when all this AI stuff started coming about it becomes very easy to follow that train of thought into thinking wow, guess they made computers conscious too like animals, but that last bit pretty much really is just marketing bullshit.

          The AI technology that exists now is pretty amazing in that it can now easily trick people into thinking it is conscious or works similar to the mind of a living organism, but it’s literally only designed to appear that way, not actually be that way. It’s all incredibly complex stuff, though. So much so that I still kinda do have hope we’ll get there eventually.