[02:19:51] <SeH_> archels: http://blog.wholebraincatalog.org/?p=65
[06:13:12] <SeH_> archels: the kurzweil talk is great
[10:23:22] <archels> Hmm, I thought that DNA mobility was not restricted to neurons... fascinating nonetheless.
[19:43:24] <archels> hey SeH_
[19:43:43] <archels> I was thinking about your claim that critters are about as smart as fish now
[19:44:01] <archels> How many neurons do these critters have, approximately?
[19:44:35] <SeH_> archels: 2000 but thats with the fucked up brainz
[19:44:46] <SeH_> i recoded the brainz and i think it can be done with far fewer
[19:44:56] <SeH_> http://crittergod.svn.sourceforge.net/viewvc/crittergod/crittergod4/src/neural/
[19:45:02] <SeH_> check out neuron.h and brain.h to see the internals
[19:45:07] <archels> I think the average fish has quite a few more neurons than 2000
[19:45:10] <SeH_> the presets are balanced postiive and negative circuits
[19:45:22] <SeH_> yes but these are not the same as physical neurons
[19:45:24] <SeH_> these operate much faster
[19:45:32] <archels> Surely you don't claim to have modeled their entire visual and sensory systems in the brain?
[19:45:34] <SeH_> thousands time fasters
[19:45:43] <SeH_> in critterding,just visual
[19:45:47] <SeH_> they have retinas
[19:46:09] <SeH_> i could make a video ... they blink in interesting ways
[19:46:16] <SeH_> steady state with metastability
[19:46:21] <SeH_> if inputs are frozen
[19:46:28] <archels> I realise that, but I would contend that you could model the entire visual system of a fish in 2000 fairly-biologically-accurate neurons, let alone the entire brain
[19:46:45] <SeH_> im not claiming that
[19:46:49] <SeH_> i said they appear as intelligent as fish
[19:46:52] <SeH_> seeking food
[19:46:54] <SeH_> is their only behavior
[19:46:58] <archels> Insects act in familiar ways to use. Maybe it's as smart as an insect.
[19:47:14] <SeH_> goldfish for example in a bowl, only needs to seek out food
[19:47:21] <SeH_> thats what the critterding critters are doing
[19:47:43] <archels> Yes, but that doesn't imply that the complexity of their brains is equal.
[19:47:48] <SeH_> no it doesnt
[19:47:59] <archels> So they're not as smart as fish.
[19:48:14] <SeH_> they appear as smart as goldfish
[19:48:28] <archels> That's a pretty relative statement.
[19:48:39] <SeH_> ok would you rather i not make it?
[19:49:49] <archels> Would I rather you didn't spend more time on CritterDing than on SpaceNet? :)
[19:51:26] <SeH_> i think theyre going to converge anyway
[19:51:32] <SeH_> im thinking about porting spacenet to c++
[19:51:48] <archels> For speed?
[19:52:32] <archels> Right, so the original premise of SpaceNet being an operating system is ejected?
[19:53:25] <SeH_> yes for speed and memory
[19:53:43] <SeH_> yes an operating system
[19:53:46] <SeH_> even more so in C++
[19:53:59] <SeH_> http://crittergod.sourceforge.net/about/
[19:54:04] <archels> Not really an operating system in the traditional, sense, then.
[19:54:15] <SeH_> like a windowing system then
[19:54:31] <archels> right
[19:54:36] <SeH_> opencog included inside
[19:54:38] <archels> Well, or something that manages applications.
[19:55:14] <archels> I mean, you're trying to improve upon GUI concepts like windows, right? 3D shapes and all.
[19:56:15] <SeH_> yes but also to make an AI OS
[19:56:18] <SeH_> semantics
[19:56:22] <SeH_> 3d etc
[19:56:26] <SeH_> it seems obvious to me
[19:56:34] <SeH_> the direction this stuff is going
[19:56:55] <archels> But like I said yesterday or so, building an AI OS is quite different from building an AI.
[19:57:14] <SeH_> http://librdf.org/
[19:57:23] <SeH_> AI OS = an interface for working with AI
[19:57:28] <SeH_> better than command lines or HTML web browsers
[19:58:35] <SeH_> i am not sure we'll ever achieve AGI. but i am sure we'll apply its concepts to make computing easier
[19:58:39] <SeH_> i hope we achieve AGI though
[19:58:44] <SeH_> http://sikuli.csail.mit.edu/
[19:59:26] <archels> sure, it'll be a couple decades, but eventually of course
[19:59:51] <archels> It's just a question of setting certain goals.
[20:00:08] <archels> What is your goal? HAL-9000?
[20:01:57] <archels> Precisely modeling a human brain?
[20:02:46] <SeH_> no there are more interesting alien brain types to explore
[20:02:53] <SeH_> especialy ones more amenable to electronic hardware we use now
[20:03:05] <SeH_> including those that fuse neural and symbolic systems
[20:03:14] <SeH_> we can get a headstart by using digital logic's symbolic abilities
[20:03:25] <SeH_> that otherwise would take a bio brain to support implicitly
[20:03:50] <SeH_> like a 100:1 efficiency improvement
[20:03:57] <SeH_> maybe 1000, etc
[20:04:58] <SeH_> http://aaai.org/ocs/index.php/FSS/FSS09/paper/view/871/1282
[20:04:58] <archels> So, using that metaphor, why not make SpaceNet that which is the symbolic system, the meta, the talking with files or with the web?
[20:05:26] <SeH_> it doesnt matter what its called or where spacenet or critterding or whatever starts and ends. the main thing is to make a real system that does certain features:
[20:05:44] <SeH_> file systems, symbolics, neural nets, life forms, web navigation, communication, etc
[20:05:55] <SeH_> im tired of using "applications"
[20:06:01] <SeH_> why is this IRC thing different from my email program
[20:06:06] <SeH_> etc
[20:07:48] <archels> Right, but some items on that list strike me as "OS" while others don't.
[20:08:06] <archels> Of course you can e.g. use neural nets to encode a part of the operation of the OS.
[20:08:12] <SeH_> then adjust ur definitions, the boundaries are shifting
[20:08:25] <archels> File systems, web navigation, communication.
[20:08:41] <archels> Symbolics, perhaps.
[20:08:44] <SeH_> archels how much do you understand about opencog?
[20:08:55] <SeH_> a lot of this is inspired by it
[20:09:05] <archels> OpenCog is a mess
[20:09:08] <SeH_> http://opencog.org/wiki/OpenCogBuntu
[20:09:25] <archels> I don't like Ben Goertzel
[20:09:26] <SeH_> not opencog as its implemented, but as its designed
[20:09:38] <SeH_> as seen on the wiki
[20:10:07] <SeH_> opencog is not a monolithic application. it is several subsystems that havent even been assembled yet
[20:10:41] <SeH_> its these subsystems when fully understood i can see how they can either make an OS (guided by human intellience) or possibly an autonomous agent
[20:10:49] <SeH_> and anything in-between
[20:11:01] <SeH_> archels are u familiar with the term 'exocortex'
[20:11:07] <SeH_> http://en.wikipedia.org/wiki/Exocortex
[20:11:11] <archels> Yes, SpaceNet can succeed where OpenCog fails.
[20:11:13] <archels> And yes, I am.
[20:11:21] <SeH_> well that defines it even better
[20:11:31] <SeH_> just an electronic layer around the neocortex
[20:11:37] <SeH_> interfacing with the web
[20:11:44] <SeH_> it oculd also be called an augmented reality OS
[20:12:10] <archels> alright, yeah, that sounds right on track
[20:12:22] <SeH_> so it can contain agents that have degrees of autonomy
[20:12:28] <SeH_> if you choose, u can let it go on autopilot
[20:12:31] <SeH_> or you can assume full control
[20:12:41] <archels> sure
[20:14:07] <archels> It still seems quite tangential to the goals of the CritterDing project.
[20:14:42] <archels> brb, food.
[20:14:43] <SeH_> well just imagine a 3D web browser panel with a 3d physical spider crawling across it
[20:14:50] <SeH_> that captures it
[20:14:57] <SeH_> or a snake the pushes a GUI button
[20:18:59] <SeH_> the OS becomes self aware through the agents it contains operating within it
[20:19:04] <SeH_> as they sense it from inside
[20:19:36] <SeH_> archels: thanks for asking these questions.  its helping me write a document ;)
[20:19:54] <archels> hehe I hope I'm not being annoying.
[20:20:01] <SeH_> no it is very helpful thanks
[20:20:10] <SeH_> i wish i could ask you similar quality ones to you
[20:20:22] <archels> You should write a paper or so though, yeah.
[20:20:26] <SeH_> your poking at places in my mind that i would never have gotten to myself
[20:20:39] <archels> haha, excellent
[20:20:51] <SeH_> well i have enough material for a book or two
[20:21:07] <archels> How does CiteSeer work, by the way... can you contribute stuff yourself?
[20:21:15] <archels> or some sort of open dir
[20:21:17] <SeH_> i dont know much about academic papers or citeseer
[20:21:26] <SeH_> theres another one called arxiv
[20:21:39] <archels> Yeah, exactly like that.
[20:21:39] <SeH_> just as long as its not ACM or IEEE O_o
[20:21:44] <archels> heh
[20:21:46] <SeH_> i hate ACM and IEEE
[20:21:48] <SeH_> requiring membership
[20:22:01] <archels> My uni gives me membership for free. :)
[20:22:26] <SeH_> thankfully i left academia.. i dont expect to return
[20:22:35] <SeH_> the internet is my university now
[20:22:36] <archels> Got any degrees?
[20:22:39] <SeH_> and people like you are my teacher
[20:22:42] <SeH_> BS comp engr
[20:22:52] <archels> oh nice
[20:23:55] <archels> Well anyway, bbl, going to have dinner.
[20:24:01] <archels> Nice talking with you.
[20:24:11] <SeH_> k

   January 22, 2010  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >