• Artificial Intelligence: Is It Possible and Is It Ethical?
    204 replies, posted
AI is SUPER ETHICAL I mean no less ethical than child birth at any rate, the creation of sentient life for the the purpose of X. X being whatever.
[QUOTE=Motherfuckers;35098939]AI is SUPER ETHICAL I mean no less ethical than child birth at any rate, the creation of sentient life for the the purpose of X. X being whatever.[/QUOTE] but what if you fuck up that wouldn't be moral
[QUOTE=crackberry;34441521]In the case of Watson, I think that the extent of what that super computer could do is find facts and put them forward. I don't think there is really any truth behind computers becoming self aware except in science fiction literature and movies.[/QUOTE] I fully agree with this. There is no way that a logic-based system like a computer could develop a "opinion" of its own. That would be like making a string of binary with a 2 in it. Its just not possible.
[QUOTE=Mr_WOBBLeS;35133628]I fully agree with this. There is no way that a logic-based system like a computer could develop a "opinion" of its own. That would be like making a string of binary with a 2 in it. Its just not possible.[/QUOTE] "There's absolutely no way a carbon-based system could develop an "opinion" of its own. That would be like making a DNA string with a transistor in it!"
I think there is an important distinction to be made between self-improving artificial intelligence and whatever the fuck you call AI that do not have that capacity. The former is significantly more dangerous.
[QUOTE=Splendor;35139784]I think there is an important distinction to be made between self-improving artificial intelligence and whatever the fuck you call AI that do not have that capacity. The former is significantly more dangerous.[/QUOTE] Exactly. Also, an AI that is significantly smarter than humans but non self-improving is still a threat as well.
[QUOTE=Splendor;35139784]I think there is an important distinction to be made between self-improving artificial intelligence and whatever the fuck you call AI that do not have that capacity. The former is significantly more dangerous.[/QUOTE] Like you fucking know? Is this seriously a thread debating shit we've heard about in movies? Reading the first 3 pages seems to say so. No group or collection of people have ever created anything resembling AI in the way that the OP describes it (a self aware being capable of learning and so on). Therefore, there are no design proposals to debate, no problems to discuss, and no actual substance to this debate at all really. If we somehow gained the ability to create an artificial (non-biological) neural network (provided of course we begin to understand neural networks to a point where we can solve complex problems with them such as the mammal brain does) that could be used in some way, we would already understand what we were building and what our machine would be capable of doing. Its not like some fucking mad scientist group is going to accidentally create a super-human like robot that can do everything better than humans unexpectedly. Step out of the cave and into reality for a split second and you'd see that. Every person who has posted here (excluding perhaps satane's post) has demonstrated a clear lack of understanding of how neural networks actually work and how you [b]could never[/b] accidentally create a human-like AI being without understanding [b]exactly[/b] what you're doing. There is also no foreseeable reason to ever want to create a human-like AI because the human brain has things that would be unnecessary to a robot designed for a specific task and again [b]could not magically evolve those capacities[/b] should they not be designed to have them. We might as well debate which religion is more correct.
[QUOTE=Slight;35140405]Like you fucking know? Is this seriously a thread debating shit we've heard about in movies? Reading the first 3 pages seems to say so. No group or collection of people have ever created anything resembling AI in the way that the OP describes it (a self aware being capable of learning and so on). Therefore, there are no design proposals to debate, no problems to discuss, and no actual substance to this debate at all really. If we somehow gained the ability to create an artificial (non-biological) neural network (provided of course we begin to understand neural networks to a point where we can solve complex problems with them such as the mammal brain does) that could be used in some way, we would already understand what we were building and what our machine would be capable of doing. Its not like some fucking mad scientist group is going to accidentally create a super-human like robot that can do everything better than humans unexpectedly. Step out of the cave and into reality for a split second and you'd see that. Every person who has posted here (excluding perhaps satane's post) has demonstrated a clear lack of understanding of how neural networks actually work and how you [b]could never[/b] accidentally create a human-like AI being without understanding [b]exactly[/b] what you're doing. There is also no foreseeable reason to ever want to create a human-like AI because the human brain has things that would be unnecessary to a robot designed for a specific task and again [b]could not magically evolve those capacities[/b] should they not be designed to have them. We might as well debate which religion is more correct.[/QUOTE] [url]http://singinst.org/summary[/url] hurr [editline]14th March 2012[/editline] it's like saying "nobody has ever created an atomic weapon before, so there's no point in worrying that it might [url=http://www.fas.org/sgp/othergov/doe/lanl/docs1/00329010.pdf]ignite the atmosphere[/url]" [editline]14th March 2012[/editline] also why are you privileging neural networks over other mind architectures? why would the mind architecture produced by evolution also happen to resemble the optimal one?
[QUOTE=DainBramageStudios;35141637][url]http://singinst.org/summary[/url] hurr [editline]14th March 2012[/editline] it's like saying "nobody has ever created an atomic weapon before, so there's no point in worrying that it might [url=http://www.fas.org/sgp/othergov/doe/lanl/docs1/00329010.pdf]ignite the atmosphere[/url]" [editline]14th March 2012[/editline] also why are you privileging neural networks over other mind architectures? why would the mind architecture produced by evolution also happen to resemble the optimal one?[/QUOTE] [QUOTE]In 1965, I.J. Good proposed that machines [B]would one day be smart enough to make themselves smarter.[/B][/QUOTE] Why should they do that?
[QUOTE=DainBramageStudios;35141637]I'm dumb also why are you privileging neural networks over other mind architectures? why would the mind architecture produced by evolution also happen to resemble the optimal one?[/QUOTE] As much as I don't want to respond to your effortless post I will out of boredom (and I the fact that I expected no less). I humored you and read the first 3 paragraphs of that website, checked the citations, and didn't look back. Why are you bringing "other mind architectures" into the debate? Those don't exist yet. So yeah, neural networks being the only "mind architecture", they are default the most optimal. I used neural networks as solutions to problems that humans face because it is the current direction of the science and I also trust the billions of years of evolution that gave me the ability to type this sentence. Any type of learning machine will be subjected to all of my points anyway. [QUOTE=matsta;35144250]Why should they do that?[/QUOTE] 90% of the statements made or referenced by that website are complete speculation and fantasy, a lot of those authors are not even qualified to be making those statements (having a phD doesn't make you magically qualified to say anything you want). How can you make predictions about AI that doesn't exist and claim results of studies on artificial systems that don't yet exist? Its fucking insane.
[QUOTE=Slight;35145620]Why are you bringing "other mind architectures" into the debate? Those don't exist yet. So yeah, neural networks being the only "mind architecture", they are default the most optimal.[/quote] ummmmmmmm no that's not how it works. think about the space of all possible minds, and then take the subset of that which satisfies these conditions: a) are buildable by humans b) are well suited for general problems, such as managing the earth's resources c) are self-aware, conscious, whatever you want to call it d) have the capability to self-modify neural networks aren't going to be in that, or even if they are, they're not going to be the best option. again, just because evolution spat them out doesn't mean we can't try to do better. neural nets are just the [B]local optimum[/B]. [quote]I used neural networks as solutions to problems that humans face because it is the current direction of the science and I also trust the billions of years of evolution that gave me the ability to type this sentence. Any type of learning machine will be subjected to all of my points anyway.[/quote] evolution isn't an upward path, holy shit [B]it's a miracle that human brains can even count. the retina is built upside down and back to front. the reccurent laryngeal nerve [I]loops around the heart.[/I][/B] I'm not saying that neural networks can't be useful - they're just not practical for everything. [quote]90% of the statements made or referenced by that website are complete speculation and fantasy, a lot of those authors are not even qualified to be making those statements (having a phD doesn't make you magically qualified to say anything you want). How can you make predictions about AI that doesn't exist and claim results of studies on artificial systems that don't yet exist? Its fucking insane.[/QUOTE] um if you actually read the citations ... [editline]15th March 2012[/editline] I can imagine some doppelganger of you, roundabout the stone age, saying that wheels were impossible to create and that we should construct needlessly complex bipedal machines. Or in 19th century, and saying that lightbulbs are poppycock and that we should use chemoluminescence because that's what nature came up with.
You all know about GLADoS, right? There is my main reason for not wanting human level AI. All the Neurotoxins...
Javik the Prothean said it best: "Organic life does not know its origins. Its purpose. But an AI knows both. We created it to serve its purpose. Given time, it will view our existence as pointless." Assuming AI like this is possible in our lifetimes, a SkyNet-type situation seems very likely.
Listen, I hate you too much to continue replying to your posts and still remain out of the refugee camp, so this is the last time I'm replying to your junk. [QUOTE=DainBramageStudios;35147663]ummmmmmmm no that's not how it works. think about the space of all possible minds, and then take the subset of that which satisfies these conditions: a) are buildable by humans b) are well suited for general problems, such as managing the earth's resources c) are self-aware, conscious, whatever you want to call it d) have the capability to self-modify neural networks aren't going to be in that, or even if they are, they're not going to be the best option. again, just because evolution spat them out doesn't mean we can't try to do better. neural nets are just the [B]local optimum[/B]. [/quote] Neural networks are all of those things fool. Neural networks created [b]everything[/b] including the computer you're using. Human's are not the ideal neural configuration obviously, we are improving as the generations go on. I fully believe we'll posses the ability to even modify our own genetics directly in the future. (We already can to some extent) You are simply wrong, again. Oh and if you were wondering why I hate you, its basically that quote up there. You put words in my mouth by saying I believe evolution has the ideal solution for everything. But only after pretending to know what neural networks capabilities are like you're a fucking expert on their use and what their limitations are. Neural networks don't have to be biological either, but it sure as hell makes creating something we don't understand yet a lot easier when you have a living working reference of it to start with. When did I ever say that neural networks are the end-all solution?? I'm saying that's where we're going right now and its a promising field for so many realms of science. If there were other things that could do what neural networks do, then YES lets talk about those, [b]but there isn't[/b], so shush. [QUOTE=DainBramageStudios;35147663] evolution isn't an upward path, holy shit [/quote] What is evolution Charles? [QUOTE=DainBramageStudios;35147663] [B]it's a miracle that human brains can even count. the retina is built upside down and back to front. the reccurent laryngeal nerve [I]loops around the heart.[/I][/B] I'm not saying that neural networks can't be useful - they're just not practical for everything. um if you actually read the citations ... [/quote] Here's 1 of many papers to contradict "your" logic: [url=http://evolution.binghamton.edu/evos/wp-content/uploads/2011/07/Laats-Bergman-Throat.pdf]recurrent laryngeal nerve is not evidence of poor design[/url] and the retina is not "built" upside down, where the hell did you go to school. Its built the exact way it needed to be built. You're acting like we're all built terribly wrong yet you don't know how we're supposed to be built. Again, insane. All of life is a miracle, just ask the Insane Clown Posse. No seriously, what's your point... So we're not calculators? So many inventions and discoveries were heavily inspired by nature I don't care to list them. Use your imagination. Nature has been the direct source of solution for many human problems. [QUOTE=DainBramageStudios;35147663] I can imagine some doppelganger of you, roundabout the stone age, saying that wheels were impossible to create and that we should construct needlessly complex bipedal machines. [/QUOTE] Oh cool, let me try. I can imagine some doppelganger of you, roundabout the stone age, speculating about the possible limitations, problems, and physics of vehicles when the wheel hasn't even been invented yet. You were a good story teller until my doppelganger clubbed you over the head and smeared feces all over your cave paintings. [QUOTE=DainBramageStudios;35147663] Or in 19th century, and saying that lightbulbs are poppycock and that we should use chemoluminescence because that's what nature came up with. [/quote] The only funny thing about these analogies is how off they are. How does this apply to what I'm saying? I'm saying be practical because you're having wet dreams about some magical "mind architectures" that don't exist. I'm not seeing a lightbulb and saying, that's dumb. I'm seeing a guy talk about a glowing glass ball of light powered by magic in a world lit by the sun and saying, that's dumb. And it is, because theres fantasizing about the future, and then there's pretending like your fantasies are real and having a serious debate about them. You're worrying about problems that don't exist about inventions that don't exist. So please continue talking about your magical glowing ball of light and keep speculating how that glowing ball of light might get TOO BRIGHT AND BURN PEOPLES EYES OUT IF WE PUT TO MUCH MAGIC INTO IT OMG.
I don't think it's immoral/unethical to create true AI. Destroying it is another story though.
I don't see why not. As long as we use it as a tool for work and other stuff, not to create new humans or something. That would just frighten me and make me scared of them taking over (hurr).
[quote]What is evolution Charles?[/quote] [url=http://lesswrong.com/lw/kt/evolutions_are_stupid_but_work_anyway/]Inefficient and stupid.[/url] [quote]Here's 1 of many papers to contradict "your" logic: [url=http://evolution.binghamton.edu/evos/wp-content/uploads/2011/07/Laats-Bergman-Throat.pdf]recurrent laryngeal nerve is not evidence of poor design[/url][/quote] i actually cannot believe that you're referencing a paper written by a [B][I][U]creationist[/U][/I][/B] [quote]and the retina is not "built" upside down, where the hell did you go to school. Its built the exact way it needed to be built. You're acting like we're all built terribly wrong yet you don't know how we're supposed to be built. Again, insane.[/quote] actually no if I hired an engineer to build an eye, and they came up with a design structured like the human eye, I'd fucking fire them on the spot. [quote=http://en.wikipedia.org/wiki/Evolution_of_the_eye#Evolutionary_baggage]The eyes of many taxa record their evolutionary history in their imperfect design. The vertebrate eye, for instance, is built "backwards and upside down", requiring "photons of light to travel through the cornea, lens, aqueous fluid, blood vessels, ganglion cells, amacrine cells, horizontal cells, and bipolar cells before they reach the light-sensitive rods and cones that transduce the light signal into neural impulses, which are then sent to the visual cortex at the back of the brain for processing into meaningful patterns."[32] The camera eyes of cephalopods, in contrast, are constructed the "right way out", with the nerves attached to the rear of the retina. This means that they do not have a blind spot. This difference may be accounted for by the origins of eyes; in cephalopods they develop as an invagination of the head surface whereas in vertebrates they originate as an extension of the brain[/quote] [url]http://lesswrong.com/lw/wk/artificial_mysterious_intelligence/[/url]
[QUOTE=DainBramageStudios;35161630]i actually cannot believe that you're referencing a paper written by a [B][I][U]creationist[/U][/I][/B][/QUOTE] there are tons of creationists out there that pay attention to scientific facts, don't generalize.
[QUOTE=Raijin;35162834]there are tons of creationists out there that pay attention to scientific facts, don't generalize.[/QUOTE] ....Pay attention maybe but do not believe them.
einstein was a creationalist to some degree. he beleived that some higher force created the universe and its laws.
[QUOTE=Robbi;35162981]....Pay attention maybe but do not believe them.[/QUOTE] You can't believe in facts. Belief implies that there's no evidence.
You've repeatedly avoided my point, so one last time so its very clear: [QUOTE=DainBramageStudios;35161630] i actually cannot believe that you're referencing a paper written by a [B][I][U]creationist[/U][/I][/B] [/quote] ad hominem + straw man = good argument right? I'm well aware he's a creationist, and yes his conclusion in that paper is a little wacky in my opinion, but he makes some good points in the paper and his references were mostly solid. I don't study this shit and I doubt you do either, so make points from your own understanding instead of linking to entire websites alone. If your point is that evolution makes mistakes, who in their right fucking mind would disagree with you? [b]That's not my point[/b]. [QUOTE=DainBramageStudios;35161630] actually no if I hired an engineer to build an eye, and they came up with a design structured like the human eye, I'd fucking fire them on the spot. [/quote] Stop with the analogies, they're literally your weakest attribute. Please find me an engineer that can create an eye to fit in a self sustaining organism and that can take [b]half[/b] the punishment that the human eye can take and function [b]half[/b] as good as the human eye. If there is no severe penalty to the eye being built upside down, [b]why does it matter?[/b]. Is that the crux of your argument? And when I said the eye isn't built upside down, I figured you where speaking more metaphorically not literally. My bad. No one here is arguing that the human eye is the most advanced/efficient eye ever. I'm simply saying, its good enough to do what it needs to do. Maybe I'm totally wrong though, I'm not going to argue that right now because it has never been my point. [QUOTE=DainBramageStudios;35161630] [url]http://lesswrong.com/lw/wk/artificial_mysterious_intelligence/[/url][/QUOTE] So basically you avoided my point completely because you realize you can only win your straw man argument? I've already stated that neural networks are not the only solution to artifical intelligence. [b][highlight][u]My Point[/u][/highlight][/b] I'm saying [b]we have not explored any avenue for AI that meets the thread's criteria so lets stop talking about problems that could exist about things that are well off in the future and any speculation about them will mostly likely be totally inaccurate[/b]. I'm also saying [b]neural networks are a [u]window[/u] into how intelligence can work and can be used as a learning tool or source of inspiration for future AI systems[/b]. They are [b]not[/b] the ideal model of intelligence necessarily but a potential aid in understanding a problem that has gone studied yet unsolved for decades. The points in [b]bold[/b] are my argument. If you choose to respond to other minute details because you think you know more than me about it, fine and maybe you do, but that's not my argument and has never been.
[QUOTE=Slight;35165244]If there is no severe penalty to the eye being built upside down, [b]why does it matter?[/b]. Is that the crux of your argument? No one here is arguing that the human eye is the most advanced/efficient eye ever. I'm simply saying, its good enough to do what it needs to do.[/quote] There is a penalty, we have a blindspot where the optic nerve threads through the tissue. Either way, wouldn't it be [I]prudent[/I] to at least get the AI design right the first time? We don't have billions of years to refine the design, it could end us in a month if we build it wrong. [quote]And when I said the eye isn't built upside down, I figured you where speaking more metaphorically not literally. My bad.[/quote] oh ok nevermind [quote]I'm saying [b]we have not explored any avenue for AI that meets the thread's criteria so lets stop talking about problems that could exist about things that are well off in the future and any speculation about them will mostly likely be totally inaccurate[/b]. I'm also saying [b]neural networks are a [u]window[/u] into how intelligence can work and can be used as a learning tool or source of inspiration for future AI systems[/b]. They are [b]not[/b] the ideal model of intelligence necessarily but a potential aid in understanding a problem that has gone studied yet unsolved for decades. The points in bold are my argument. If you choose to respond to other minute details because you think you know more than me about it, fine and maybe you do, but that's not my argument and has never been.[/QUOTE] Ah ok that's fine. I was under the impression that you were for building a greater-than-human intelligence with a similar design to human minds. Neural networks do give insights into how intelligence works, including our own. [editline]16th March 2012[/editline] [QUOTE=Satane;35163406]einstein was a creationalist to some degree. he beleived that some higher force created the universe and its laws.[/QUOTE] No, when Einstein talked about "God" he was referring to the underlying structure of the Universe, and the sense of wonder he felt, not a deity. When he said stuff like "God does not play dice", it was shorthand for him saying that he disagreed that the Universe was nondeterministic. It was a horrible mistake in phrasing on his part, and it's now been immortalised by creationists trying to get Einstein on their side. He was in no way a creationist.
[QUOTE=Satane;35163406]einstein was a creationalist to some degree. he beleived that some higher force created the universe and its laws.[/QUOTE] Completely false, he believed in Baruch Spinoza's abstract conception of god which is equivalent to the order of nature.
I'm thinking that maybe the only way to truly understand how the human mind or intelligence works in general is by attempting to model it. The human mind is composed of systems and those systems are composed of neurons; aren't neurons basically like organic circuitry? By modeling that circuitry and those systems with our computer processing I believe that we can emulate a human mind, and by doing that we can perhaps understand our thought processes and why we do what we do.
Sorry, you need to Log In to post a reply to this thread.