The latest news/musings from him were about a fortnight ago.
Before I forget: Congratulations to neurobot for getting his PhD! And good luck to Magos with the Master's. And Ettina, we need to talk, to make sure you don't get kidnapped by the wrong kind of AI people. It's fun hearing about how you're all getting on.
Me? Oh, well, you know. I'm programming and stuff. I'm still working on the details for my generalized cortical map. It's very slow work but it feels like it's steadily coming together. I have to remind myself every now and then that I'm trying to create something that's alive and able to think for itself, because it's easy to lose sight of that in all the technical minutiae. Code to prioritize subliminal and supraliminal yang signals or create fractal noise in motor sequences kind of gets to be an end in itself, and I forget that what I'm really doing is giving a little creature the ability to play and to have dreams and hopes.
I do hope this is all going to work out like I want it to. It's still so hard to tell, when they can't even walk yet. Who knows whether these creatures or their offspring will ever be able to climb above the level of totally stupid. But at least they're real. Everything they do is an emergent consequence of a set of principles that I think, even more than I did when I started, might explain something important about what's going on in our own heads to create our own minds. I'll settle for stupid but conscious over smart automatons, I reckon!
Talking of play, here's a thing you can maybe help me with. The importance of play has been becoming ever more obvious. My design currently has three main kinds of learning going on inside each of the maps that make up the brain, and they all involve learning by observation. [Edit: Now that I think about it, there are five.] Which means there has to be something to observe, and there's only so much you can learn by passively watching the world go by. We learn by playing around with stuff - with our bodies to begin with, and then as we learn how to get our bodies to do useful things, we use them to play around with objects and people and behaviors.
Play sort of starts on the outside and works in, so newborn babies are constantly just flinging their arms and legs around to see what they do. But as they get a grip on how moving their arm like this allows them to do that, they gradually start focusing their attention more narrowly and toying around more deliberately. By the time you get to my great age you're learning how to fly 747's for fun or playing with code. But back in those early days, once the whole-body waving of limbs has been brought under control, babies still tend to focus on one thing at a time. What does this finger taste like? Look how funny it is when I drop momma's necklace down the back of the sofa, etc. In my terms, this basically means the creatures should be selective about which particular cortical map to experiment with at any moment - what happens if I shift my balance in random ways, or what can I do with my left arm, or I wonder what's over there? So play is a top-down activity - we decide centrally to do it and then we focus our attention on playing around with just one limb or object or category of behavior at a time.
This makes sense - if we flail everything around at random we'll never be able to figure out what is affecting what. And later play involves hand-eye coordination and the focusing of multiple sensory systems onto the same thing. But WHEN do we play? When do we practise skills? How do we choose what to play around with? This sounds pretty obvious when I say it out loud, and yet I can't quite get a handle on what the rules should be. Given that at least two of these three learning mechanisms involve trying stuff out (completely or partially at random) it's really critical. But when should a baby creature wave its arms around to see what they do? Not when something important is going on, for sure. At those times we tend to stick to what we know, even if it doesn't work. Play is for when we're relaxed. But it could look kind of odd without some rationale. I'm worried that in order to get a creature to learn, it will have to just decide at random to experiment one moment with twitching its left leg or another with exploring the possibilities of head movements, and the result is going to look like it has severe Tourettes Syndrome or even epilepsy. There needs to be more logic to it than that. Discovery is obviously important, for instance - a baby will notice its own foot moving and then stick it in its mouth to see what happens. If it doesn't see the foot then it won't do that, so there's some logic to it. But that's only part of the story.
Since you don't really know yet what all the cortical maps do (nor do I, for that matter) it's probably hard to give me any insights. But I'm just wondering if anyone has anything in general to offer on the question of when babies feel playful, what it is that delights them, and how they choose between things or body parts to play with. Just when is the right time to try sticking a penny up your nose?. Any general insights into, or anecdotes about, play, practice and trial-and-error, particularly in babies, would be great to hear. Thanks.
Author
Topic: Grandroids: Secret Supporter Stuff! (Read 65882 times)