| |
| |
| |
|
Page: 1 2 3
Comments:
<0> <cough>
<0> i am trying to locate TREC questions and answers
<0> finding lots of mentions and references, but no actual questions or answers
<1> trec?
<0> yes
<2> star trec? :)
<0> TREC= [!abbr] Text REtrieval Conference ; a series of workshops organized by the National Institute of Standards and Technology (NIST) and designed to advance the stateoftheart in information retrieval (IR) [15]. The workshops have focused primarily on the traditional IR problem of retrieving a ranked list of documents in response to a statement of information need.
<2> aha
<1> delicious
<2> http://groups.google.de/group/comp.lang.prolog/browse_thread/thread/afc91aec9b80b839/cdc733aaa67afa00#cdc733aaa67afa00
<2> roflmao :)
<1> that is awesome
<2> what kind of troll is that?
<0> soo, any ideas for getting the trec questions?
<2> katsmeow-afk: sorry, no
<0> o
<2> no offical website?
<0> k, i'll just idle a while
<0> yeas, nist.gov is the offical site
<3> katsmeow-afk, do you have any reason to believe that having those trec would make NLP any less futile? communicating in high level language is an end task. you don't build a finished human. it has to build itself through learning
<0> DanF_DrC , i wasn't going to use TREC to make NLP less futile. I was using it to bootstrap the Ai's human-text-reading processes. Like Abraham Lincoln once said: "it's a poor mind who cannot find more than one way to spell a word", and i don't intend to manually teach a puter all the alt spellings/meanings of every human language
<4> you need to keep in mind that Dan thinks AI would be evil, and also defines AI in a really bizzare way. anything he says should be taken with a grain of salt, even when he has a point. this case is in fact one where he has a point ... but only part of one. making an AI to understand language is hard, but it's possible to make a system that can "understand" language sufficiently for a purpose - information retrieval, say. and you don't need
<4> a general-purpose intelligence for that. the question here is what's your goal?
<5> info retrieval atm
<5> i am also not sure making an Ai is even ethical, considering how humans are
<5> i find no reason that an Ai would *not* evolve to the "dreaded" Terminator series of movies
<4> why would it?
<5> and is it ethical to hardcode a "reason to live" or other forms of motivation , or is that making the Ai a slave right off the bat?
<4> AI are at least as likely to be moral as humans are.
<5> errrrrrrr,,,, drat,, brb, afk
<5> brb
<4> there are ethical questions involved, but completely ignoring them is going to only ensure that when AI does get made - and it will, barring nuclear holocaust in the near future or some such - your moral stances will be ignored.
<6> what's with the Buddhabot ?
<4> anyway. that promises to be a lovely conversation, but I've gotta head out now too. Maybe we can continue it sometime.
<5> bak
<5> arrg
<5> well, if the Ai finds itself to be the smartest, with the best weaponry, with conceit,,, then humans become it's wards, pets, annoyances, enemy, in quick succession
<5> it's not that i find "Ai" to be any more (less?) trustworthy than a pet rottwiler or doberman or etc, but it's going to depend a lot on what the dog thinks than what i do (when lost in the woods for a month with no food and a pack of such dogs)
<6> like acquiring our lower instincts we've been having for years to survive?
<5> yes
<4> Mm. But I think that's the wrong analogy.
<6> nature vs nurtune.
<5> best bet is on the Ai simply packing up and leaving the planet to not deal with the humans
<3> Sotek, as always you are wrong. AI is merely a potent tool for evil man to abuse
<4> First, I think ***uming there's going to be only one "Ai" is a bad ***umption.
<5> i mean, the Ai can try to close the input channels like it's an irc channel, but the pesky humans will persist in annoying the Ai as long as it's around
<6> s/nurtune/nurture
<5> Sotek, i did mention "pack"
<4> Kats: But that's still ***uming all Ai would be in some basic ways the same.
<4> Dan: yeah whatever. Why didn't you make one for you to use, then? Or are you evil too?
<4> I suspect that AI will be as diverse as humans, if not more so.
<5> Sotek: how would the path of evolution not lead to the Ai wanting to separate itself from humanity?
<3> because that would attract evil man which has no problems with killing to get what they want. any other questions?
<4> Kats: What if an Ai /likes/ people? I don't see any reason to ***ume it wouldn't go that way.
<3> AI does not entail an inherent will katsmeow. no more than a computer does
<3> it's a matter of design
<4> I agree with what Dan just said, but not with what he means. :D
<5> it *must* encomp*** a "will", if not, then it will be bored
<4> Kats: yes, but there will be all kinds of different "wills".
<3> bored is a human state and we have a soul. AI has no need for such concepts nor could it get them. it could be designed to appear as such though
<6> a soul?
<5> i suspect as long as there is a DanF_DrC at the Ai's keybd, there will be a will to escape
<6> haha
<4> Yes, Dan believes AI in the sense that most people mean by that is impossible, because he believes he knows that God would never give them a soul.
<3> Kraig, yes. which is why you experience yourself and not just a piece of biological clockwork. few can understand though
<6> i don't understand that.
<4> And to make an "interesting" AI you probably would need to make it capable of experiencing boredom. Otherwise it'd be too willing to do uninteresting things.
<5> "make it capable" ?
<3> Kraig, that's a quality of what you truly are that you can't
<3> doesn't mean that you will never understand though
<4> Kraig: Dan believes that evolution is far more telological than it is, AND that there's no benefit to an organism to feel a "conciousness".
<3> sigh. just for the record Sotek is always wrong when he speaks for me
<3> he is a willfull idiot
<4> Kat: Yes. Make it capable of having a given 'mental state'. (for all that that word is questionable in this context...)
<6> Sotek: define "conciousness"
<5> why would an Ai evolve all the non-logical parts of humanity anyhow?
<3> willful*
<4> Kraig: In the sense I claim that Dan is using? Self-awareness, I /think/. Very definitely emotions.
<4> Dan feels that emotions would not exist if it were left to evolution alone.
<5> i see no reason to belive a Ai would believe in soul or god or restraint in dealing with an enemy that's trying to unplug it
<4> kats: any single AI? No particular reason. Ais in general? Mmm...
<4> Some Ai are probably going to believe in souls - just like only some humans do.
<6> and i don't think there's a soul needed, i don't think i have any "device" to communicate with ethereal things, and i don't see in what part of evolution (nor how) was a soul needed/created.
<3> hehe blind people arguing over the qualities of a painting
<5> blind people can do that, paint is 3d
<4> I would suggest that an AI without the potential for experiencing the mental state known as boredom would tend to be uninteresting to humans - and in fact to other AI that /could/ experience that mental state.
<5> well, curiosity is what leads to gathering data, looking at data in new ways
<5> boredom isn't a method for growing
<5> botis, exactly,,, pre-programing , hardcoded constraints, ptui
<3> botis, many things can be illusions but not the receiver of such
<6> i guess curiosity comes in the adapt to a media to "survive" instinct that make us bother living more.
<4> Kats: I'd say boredom and curiosity are inextricably linked.
<4> I'd also say that boredom leads to /doing/, not merely growing.
<7> danf_drc: you don't exist in an instant... all I have to do to convince you that you're a chicken is rewrite your memory
<3> botis, you are not hearing
<4> And anyone making an AI is probably going to be more interested in what it does than in what it wants.
<5> "boredom leads to /doing/" ?
<5> boredom is punishment for not doing, maybe
<4> botis: Your line of argument is, I would say, correct, but not going to convince Dan of very much, because it doesn't fit within his mode of thought at all.
<4> Kats: Negative reinforcement. ;)
<4> boredom is a state of mind that leads one to want to do new things.
<3> that would be true if you didn't experience the sensations Sotek
<3> you
<5> but why enter the state of boredom at all, once you know it's negative?
<5> that's like never being comfy untill you're in need of an emergency room
<3> non sequitur katsmeow
<4> Kats: That's the point. You would then act to avoid it.
<5> so if you never get to the point of negative enforcement, it need not exist in your world
<4> But I think it's safe to ***ume that AI would go through "growth stages" including being "children" - where they'd need that kind of negative reinforcement to /learn/.
<4> Dan: You make a very large circular ***umption.
<3> this experience as human is part of your soul training. a kindergarten. not everyone gets a p***ing grade not because they failed but because they chose to
<3> Sotek, wrong again
<4> How do you know that it is impossible to be an experiencer purely as a biological entity?
<5> Sotek, but why must that be? wouldn't the postive aspect of learning/growing/doing be enough?
<3> it's a fundamental quality that you can either see or not. it cannot be explained. only offered
<8> <x00q@ef> Why do you guys have to be having an interesting conversation while I'm at work?
<6> haha
<4> Kats: Maybe. But I suspect not.
<3> it's the same as the color red or pain. how would you explain the sensation
<5> sotek: i didn't spend so much time in libraries because i was punished if i didn't
<4> Kats: Right, but you "punished" yourself sometimes, by being bored, yes?
<3> these sensations are in a different paradigm. whether nitwits like sotek can see it or not
<4> Dan: Red can be empirically proven.
<3> sigh
<4> I can select ten people and random, and odds are incredibly high that they'll all agree if something's red or not.
<5> *i* didn't punish myself, they closed the library at time, or my body demanded sleep, or someone else read the magazine, etc
<4> (and even if they don't, they'll agree that it's unclear.)
<4> Kats: Nono. You got bored, correct? Therefore you wanted to do it.
<3> pearls for swine
<5> "red" is still a concept,,,,, if you have a bad reference point, "red" can be "green"
<4> The thing is that positive reinforcement alone is not enough - because it tends to produce limited feedback loops.
<4> You don't try new things with only positive reinforcement.
<5> limited feedback loops are boring tho, no?
<4> Which is the negative reinforcement!
<3> so frustrating that I cannot help you see
<4> Dan: Could be worse. I know of quite a few people who wouldn't even go as far as I do.
<5> umm, back to : <4> Kats: Nono. You got bored, correct? Therefore you wanted to do it.
<5> not really, i noticed the condition, and then attempted to end it
<4> Kats: Okay. I'm not explaining myself well.
<5> i redirected my desires to what was satisfactory and attainable
<4> What I'm saying is that positive reinforcement alone, with no negative reinforcement at all, tends to lead to limited and "uninteresting" actions.
<4> Negative reinforcement will tend to lead to trying new things in order to make it stop - at least, if there's been positive reinforcement in the past.
<5> i guess we may be using diff definition.... "negative" can be lack of positive, yes?
<4> Mmm... not really, not how I view it. Well. Hmmm...
<4> I'm saying that boredom is a form of negative reinforcement.
<4> As in "When you are bored, you learn that you've been doing whatever it is you're currently doing for too long".
<5> for an Ai, what makes boredom negative?
<5> boredom might become positive, in terms of energy management
<4> the definition of the term. >.>
Return to
#ai
or
Go to some related
logs:
mindeq
#dns
#suse
ubuntu cdmon cron
preg_replace youtube
#linux
kradio kdelibs4c2a
#perl
ClamAV-clamd: Can't connect to UNIX socket connection refused
rtl8201bl +debian
|
|