cuts and unicode characters?

[alexr_rwx]

If I was going to despair about the state of things, I'd say:
Machine-learning is ridiculously hard, and what little I can wrap my puny brain around, there are cleverer people out there who are going to understand it better, and they'll be the ones that come up with clever new things. If I was going anywhere with this, I'd be a soviet mathematician from the early eighties, or possibly Indian.

Symbolic-style AI is a bit of a crock. Nobody honestly thinks that brains are rule engines (it's pretty clear that we don't go around formal-logicking everything out), and the rule engines that we can make are so brittle and the problems we can solve with them are such toys as to be trivial and uninteresting. Nobody /really/ cares about truth-maintenance systems. The AI Winter happened for a reason.

Cognitive science. (there, that was the joke). Our understanding of thought and internal representations is in the dark ages. You're not going to get at any of it without cutting open some heads, and that's questionable. Where and how do you keep the memories of your grandmother? Does anybody even have an idea how to address that problem? Does anybody care what it means to be having an experience, or is this going to be Searle and Dennett over in the corner dogmatizing at each other for forever? "Subjectivity", wtf?



If I was going to be upbeat about it, I'd say:
So I guess I'd better work hard and hope for some insight then, yes? Or maybe even if I don't make any fundamental breakthroughs, I can have fun messing around?

Comments

[eponis.livejournal.com]

A human brain is a collection of little squishy globlets, connected to each other by tendrils, that send electrical signals back and forth by a simple process. Easy to replicate.

However, a tadpole brain is the same thing on a vastly smaller level. I think you can easily imitate a tadpole brain.

Thus, I conclude that you can do this - indeed, *will* do this. It'll just take time. By the end of our generation, though, it'll be amazing what we've done.

Randomly, have I made you play around with this (http://www.ifiction.org/games/play.phpz?cat=46&game=369&mode=html)? You might find it interesting.

(I did see an excellent lobster simulation, recently, but it seemed pretty rigged.)

[oniugnip.livejournal.com]

Mmm... well, I can't imitate a tadpole brain, personally... not yet, anyway. As I understand the current level of understanding with how real neurons work. We have the artificial neural nets, of course, but it's not clear that those meaningfully do the same thing as real neurons -- and is that the whole story with how brainful things work? S'not clear. (Conscious Entities (http://consciousentities.com) often has really good discussions about this sort of thing).

But yes: maybe-hopefully :) Beautiful things, coming right up!

Also: Ooh, a thing... *fiddles*