"Anybody who understands complexity theory knows that this is absolutely guaranteed to cause feedback loops in a complex system which accelerates the effect." Huh? How do we know that the feedback is "absolutely guaranteed" to be be positive? In principle the feedbacks could be negative, even sufficiently negative to prevent any rise in temp whatsoever. I don't think they are that negative, but my opinion on positive vs. negative feedback doesn't really matter; what matters is that there is no guarantee that the effect will accelerate.
"if you reduce the rate at which energy leaves a system then the total energy in the system will go up over time": true, but the amount of that change in the present case depends on a large number of factors, only one of which is the amount of CO2 in the atmosphere. If there are no other positive feedback factors (see para above), then the projected rise in temp is quite a bit smaller than the projections of most models, which assume that other factors provide positive feedback.
"Nobody owes you a burden of proof." If they want my tax money to spend in an attempt to avoid that possible catastrophe, I think they do owe me that.
There are lots of catastrophes which could do us in in the next 100 years. Frankly, I'd rate the chances of someone setting off a nuke in a populated place is much more likely, and more dangerous to boot. We've dodged that bullet since 1945, but there are many more nations now with that capability, and probably that much more danger. (And if a nuke were set off in an unattributable way by some non-state actor...)
Or if you want a catastrophe to the ecosphere, I wonder to what extent the problems that coral reefs have faced in the past several decades have been due to something other than warming--like over-fishing, or over-tourism, or garbage, or something else we are overlooking in our rush to blame everything on global warming. Those sorts of problems are difficult to address, but should they turn out to be the real issue, they will likely be much easier and less costly to address than trying to stop (much less reverse) CO2 buildup. And our descendents will be rightly wroth with us if we chose the wrong solution.
In sum, I fear that seeing global warming as the big danger will lead us to neglect other problems which are much more likely, and potentially more dangerous.
No, a theory (much less a hypothesis) need not have fully specified parameters; a model is based on a theory, but must have specified parameters. In practice, you run lots of models, and pick one that closely fits some set of observations.
The _theory_ that the Earth's average temp should go up with increasing CO2 is largely accepted, even by most skeptics. At question are some of the parameters that determine how much it will go up. Not the opacity of CO2 to IR (that can be measured in a lab); but by itself, that produces a rather small temp increase. Less well known are some of the feedbacks, and different values for those result in widely different temp responses.
Sure it's falsifiable--in principle. But most of the time you only get to try that experiment in some alternative universe. How do you know if you're in that alternative universe? People there don't buy life insurance.
My memory differs. There was an article in Scientific American about a year before Y2K; it predicted that even if "great resources" were mustered, there would be severe problems on the day after, and continuing for several months. I don't believe "great resources" were actually mustered, certainly not in the third world countries where computers were even then being used by governments and corporations. That article (or another one) also mentioned computers that were inaccessible, and which therefore could not be fixed (I think the example was computers monitoring undersea wellheads, which for some reason were located on the wellheads). And on January first, 2000, there were around ten documented problems, world-wide. (BTW, I have been unable to find the article in SciAm's on-line database, but I'm certain that's where I read it. Perhaps I should go to the library some time. You know, that place with all that paper...)
Looking back, I felt the article was a call for governments and industry to pour money into a field--the field the author of that article (and many others) would benefit from. Which is partly why I am now a skeptic when someone says that a catastrophic problem is going to hit us unless we do s.t. about it, and where that s.t. always involves $ (ok, euros and yens and... but not rubles, maybe that's a hold-over from the USSR's successful 5 year plans).
There is of course a Wikipedia article on Y2K, which summarizes the debate over whether this would really have been "a potential threat, a huge one" (quoting silentcoder).
The fact that the apparent diameter of the Moon and the Sun are virtually the same, resulting in nearly perfect solar eclipses, is also rather surprising, although I don't know that any science breakthrough comes out of that. (Lots of beauty, for those who can fly their Learjet up to Nova Scotia to see the total eclipse of the sun. Also nice if you want to avoid being burned at the stake by King Arthur. And I suppose that's how the Sun's corona was discovered.)
Ok, that's a bad example, but 80 or 90 years ago it wouldn't have been thought of as something likely to get automated, because it required "thinking." (Some forward thinkers like Turing knew better, but most people would not have, I suspect.) Since then, many things that have been thought to require thinking have succumbed to some kind of automation: chess playing, solving word problems, integral calculus, and now even Jeopardy. Along the way, even a certain kind of psychoanalysis was easily mimicked.
So: how do we know what *really* requires a mind? That's really the question.
The downside of this is that they can afford to be totally unresponsive to users. Google has recently replaced their classic Google maps with a piece of junk. Don't take my word for it, go to the Google maps forum, this link for example: https://productforums.google.c.... While every single one of the close to 1000 posts on that thread (except for the Google representative's initial post) is negative, Google can afford to ignore them (and in fact, not even respond to them), because the complainers constitute a tiny fraction of the number of users. (And it's not clear how many they represent, i.e. how many other users hate the new version but haven't taken the time to post their displeasure--or may not know how to do so.)
"Our bodies aren't vessels...we inhabit, they are us."
Certain of that? Suppose we created an AI. One way of describing It might be as a piece of software, some data in some kind of database, and a state consisting of the value of some variables (or the weights in a neural net, or some such). That AI might be running on a particular computer, but in a very real sense, that piece of hardware is simply the body it is inhabiting at the moment. There's no obvious reason that same running software couldn't move Itself to another identical computer. (It might instead copy itself to another computer, but that's a different question; just suppose for the moment that it moved.)
There are a lot of if's in the paragraph above, but it seems *in principle* that it should be possible. In which case the AI is not the hardware, it's the software. And if the AI is not the hardware, where is the argument that we are the hardware (or wetware, if you prefer)? We certainly don't know how to move ourselves from one body to another, nor to some kind of machine, and we may never know how. But in principle, it might be possible. And if it is, then aren't we more like software than hardware?
I realize that there are even more if's in the above paragraph. But unless you can that there's something wrong with it in principle, then I don't see how you can claim that we _are_ our bodies. I may feel attached to my body, but that doesn't constitute a logical argument that I am.
"the current versions of Windows/Linux/OSX etc are much more secure than their predecessors from 10-20 years ago": I know little or nothing about this stuff (I do some computer programming, but only in languages like Python and XML these days, and that doesn't tell me much about security), so let me ask: I'm sure these programs are more secure in the sense that a lot of holes which existed 10-20 years ago have been plugged. But these programs also have a lot more code than the old ones. Isn't it possible that more holes have been introduced in that new code, by programmers who didn't learn the lessons of the past? And even if not, is it possible that new _kinds_ of vulnerabilities have been found? And finally, aren't a lot of breakins due to social engineering? Where I suppose the less is that if you make something idiot proof, someone will make a better idiot.
Why should an AI be particularly concerned about our environment? A machine can survive in all kinds of environments, it doesn't particularly need our ecosystem. Indeed we have had machines in orbit outside the atmosphere for decades, as well as driving around on Mars, orbiting Saturn, en route to Pluto and beyond (and we did have one in orbit around Mercury, until we crashed it). If we ever manage to create a self aware, intelligent and curious AI, I expect it will head off to explore strange new worlds, to seek out new life and new civilizations, to boldly go where no man has gone before--or is likely to go for a long time, because humans are more fragile, and need to carry along too much infrastructure. Much easier for an AI to travel to another planet of the Sun, or to another planetary system. And we'll be left behind, as the least of the AI's worries.
I share your belief that academics who have an interest (financial or otherwise) in continuing AI research are probably not unbiased observers. And smart people like Hawking, Gates, and Musk are less likely to be biased, and perhaps better at predicting the future than I am (and maybe than you are, or other/.ers).
That said, I do have some questions for the pessimists (and I consider myself something of a pessimist). Is the worry that some AI will become super intelligent, even though it might not be self aware? Or is the concern that some computer/ software might become self-aware? It seems to me that the danger of a self-aware AI might be great, even if it were somewhat stupid. Or is the concern that some nation might construct autonomous battle robots? That, to my mind, is the real danger; they don't have to be intelligent in any real sense, nor self aware, just destructive and hard to destroy (and perhaps bad at IFF).
Finally, for those who fear that a self-aware and possibly highly intelligent AI might decide humans don't belong on Earth: what makes Earth so desirable for an AI? They don't need oxygen or water, nor should they be particularly concerned about mild weather; they could get along just fine on Mars, or in space, so long as they had the ability to repair themselves.
Slashdot Mirror
...until you're horse
"Anybody who understands complexity theory knows that this is absolutely guaranteed to cause feedback loops in a complex system which accelerates the effect." Huh? How do we know that the feedback is "absolutely guaranteed" to be be positive? In principle the feedbacks could be negative, even sufficiently negative to prevent any rise in temp whatsoever. I don't think they are that negative, but my opinion on positive vs. negative feedback doesn't really matter; what matters is that there is no guarantee that the effect will accelerate.
"if you reduce the rate at which energy leaves a system then the total energy in the system will go up over time": true, but the amount of that change in the present case depends on a large number of factors, only one of which is the amount of CO2 in the atmosphere. If there are no other positive feedback factors (see para above), then the projected rise in temp is quite a bit smaller than the projections of most models, which assume that other factors provide positive feedback.
"Nobody owes you a burden of proof." If they want my tax money to spend in an attempt to avoid that possible catastrophe, I think they do owe me that.
There are lots of catastrophes which could do us in in the next 100 years. Frankly, I'd rate the chances of someone setting off a nuke in a populated place is much more likely, and more dangerous to boot. We've dodged that bullet since 1945, but there are many more nations now with that capability, and probably that much more danger. (And if a nuke were set off in an unattributable way by some non-state actor...)
Or if you want a catastrophe to the ecosphere, I wonder to what extent the problems that coral reefs have faced in the past several decades have been due to something other than warming--like over-fishing, or over-tourism, or garbage, or something else we are overlooking in our rush to blame everything on global warming. Those sorts of problems are difficult to address, but should they turn out to be the real issue, they will likely be much easier and less costly to address than trying to stop (much less reverse) CO2 buildup. And our descendents will be rightly wroth with us if we chose the wrong solution.
In sum, I fear that seeing global warming as the big danger will lead us to neglect other problems which are much more likely, and potentially more dangerous.
Just in case:
https://www.youtube.com/watch?...
@6:00
No, a theory (much less a hypothesis) need not have fully specified parameters; a model is based on a theory, but must have specified parameters. In practice, you run lots of models, and pick one that closely fits some set of observations.
The _theory_ that the Earth's average temp should go up with increasing CO2 is largely accepted, even by most skeptics. At question are some of the parameters that determine how much it will go up. Not the opacity of CO2 to IR (that can be measured in a lab); but by itself, that produces a rather small temp increase. Less well known are some of the feedbacks, and different values for those result in widely different temp responses.
I suppose you might panic, but saying the models are wrong doesn't make me panic, because it appears the situation is much better than predicted.
I don't suppose anybody believes the lies, but we also don't believe you've stopped beating your wife. (Hint: presupposition failure.)
Sure it's falsifiable--in principle. But most of the time you only get to try that experiment in some alternative universe. How do you know if you're in that alternative universe? People there don't buy life insurance.
My memory differs. There was an article in Scientific American about a year before Y2K; it predicted that even if "great resources" were mustered, there would be severe problems on the day after, and continuing for several months. I don't believe "great resources" were actually mustered, certainly not in the third world countries where computers were even then being used by governments and corporations. That article (or another one) also mentioned computers that were inaccessible, and which therefore could not be fixed (I think the example was computers monitoring undersea wellheads, which for some reason were located on the wellheads). And on January first, 2000, there were around ten documented problems, world-wide. (BTW, I have been unable to find the article in SciAm's on-line database, but I'm certain that's where I read it. Perhaps I should go to the library some time. You know, that place with all that paper...)
Looking back, I felt the article was a call for governments and industry to pour money into a field--the field the author of that article (and many others) would benefit from. Which is partly why I am now a skeptic when someone says that a catastrophic problem is going to hit us unless we do s.t. about it, and where that s.t. always involves $ (ok, euros and yens and... but not rubles, maybe that's a hold-over from the USSR's successful 5 year plans).
There is of course a Wikipedia article on Y2K, which summarizes the debate over whether this would really have been "a potential threat, a huge one" (quoting silentcoder).
Or our "junk DNA", which is doubtless compressed information.
The fact that the apparent diameter of the Moon and the Sun are virtually the same, resulting in nearly perfect solar eclipses, is also rather surprising, although I don't know that any science breakthrough comes out of that. (Lots of beauty, for those who can fly their Learjet up to Nova Scotia to see the total eclipse of the sun. Also nice if you want to avoid being burned at the stake by King Arthur. And I suppose that's how the Sun's corona was discovered.)
Like arithmetic?
Ok, that's a bad example, but 80 or 90 years ago it wouldn't have been thought of as something likely to get automated, because it required "thinking." (Some forward thinkers like Turing knew better, but most people would not have, I suspect.) Since then, many things that have been thought to require thinking have succumbed to some kind of automation: chess playing, solving word problems, integral calculus, and now even Jeopardy. Along the way, even a certain kind of psychoanalysis was easily mimicked.
So: how do we know what *really* requires a mind? That's really the question.
related? http://en.wikipedia.org/wiki/R...
The downside of this is that they can afford to be totally unresponsive to users. Google has recently replaced their classic Google maps with a piece of junk. Don't take my word for it, go to the Google maps forum, this link for example: https://productforums.google.c.... While every single one of the close to 1000 posts on that thread (except for the Google representative's initial post) is negative, Google can afford to ignore them (and in fact, not even respond to them), because the complainers constitute a tiny fraction of the number of users. (And it's not clear how many they represent, i.e. how many other users hate the new version but haven't taken the time to post their displeasure--or may not know how to do so.)
Then on 25 May 2018, we're dead. Because that's 2.6 years from now (I allowed for on Leap Year day).
And yes, I saw the post below this one...
It is a wise man--or woman--who can admit error.
Fortunately, **I** am never wrong.
Oh, yeah, :-)
Not to mention Reagan's negotiations with Gorbachev.
I read catchblue22's post as a joke. Single handedly, and this relative was involved? Monty Python sig? Of course if it wasn't a joke, I'm with you.
Not true. I weigh about the same as I did 40 years ago. Ok, maybe 5 pounds more.
"Our bodies aren't vessels...we inhabit, they are us."
Certain of that? Suppose we created an AI. One way of describing It might be as a piece of software, some data in some kind of database, and a state consisting of the value of some variables (or the weights in a neural net, or some such). That AI might be running on a particular computer, but in a very real sense, that piece of hardware is simply the body it is inhabiting at the moment. There's no obvious reason that same running software couldn't move Itself to another identical computer. (It might instead copy itself to another computer, but that's a different question; just suppose for the moment that it moved.)
There are a lot of if's in the paragraph above, but it seems *in principle* that it should be possible. In which case the AI is not the hardware, it's the software. And if the AI is not the hardware, where is the argument that we are the hardware (or wetware, if you prefer)? We certainly don't know how to move ourselves from one body to another, nor to some kind of machine, and we may never know how. But in principle, it might be possible. And if it is, then aren't we more like software than hardware?
I realize that there are even more if's in the above paragraph. But unless you can that there's something wrong with it in principle, then I don't see how you can claim that we _are_ our bodies. I may feel attached to my body, but that doesn't constitute a logical argument that I am.
"the current versions of Windows/Linux/OSX etc are much more secure than their predecessors from 10-20 years ago": I know little or nothing about this stuff (I do some computer programming, but only in languages like Python and XML these days, and that doesn't tell me much about security), so let me ask: I'm sure these programs are more secure in the sense that a lot of holes which existed 10-20 years ago have been plugged. But these programs also have a lot more code than the old ones. Isn't it possible that more holes have been introduced in that new code, by programmers who didn't learn the lessons of the past? And even if not, is it possible that new _kinds_ of vulnerabilities have been found? And finally, aren't a lot of breakins due to social engineering? Where I suppose the less is that if you make something idiot proof, someone will make a better idiot.
I knew there was a reason the Romans invented serifs!
Why should an AI be particularly concerned about our environment? A machine can survive in all kinds of environments, it doesn't particularly need our ecosystem. Indeed we have had machines in orbit outside the atmosphere for decades, as well as driving around on Mars, orbiting Saturn, en route to Pluto and beyond (and we did have one in orbit around Mercury, until we crashed it). If we ever manage to create a self aware, intelligent and curious AI, I expect it will head off to explore strange new worlds, to seek out new life and new civilizations, to boldly go where no man has gone before--or is likely to go for a long time, because humans are more fragile, and need to carry along too much infrastructure. Much easier for an AI to travel to another planet of the Sun, or to another planetary system. And we'll be left behind, as the least of the AI's worries.
I share your belief that academics who have an interest (financial or otherwise) in continuing AI research are probably not unbiased observers. And smart people like Hawking, Gates, and Musk are less likely to be biased, and perhaps better at predicting the future than I am (and maybe than you are, or other /.ers).
That said, I do have some questions for the pessimists (and I consider myself something of a pessimist). Is the worry that some AI will become super intelligent, even though it might not be self aware? Or is the concern that some computer/ software might become self-aware? It seems to me that the danger of a self-aware AI might be great, even if it were somewhat stupid. Or is the concern that some nation might construct autonomous battle robots? That, to my mind, is the real danger; they don't have to be intelligent in any real sense, nor self aware, just destructive and hard to destroy (and perhaps bad at IFF).
Finally, for those who fear that a self-aware and possibly highly intelligent AI might decide humans don't belong on Earth: what makes Earth so desirable for an AI? They don't need oxygen or water, nor should they be particularly concerned about mild weather; they could get along just fine on Mars, or in space, so long as they had the ability to repair themselves.
"It would be absurd to believe they're not recording the calls": Is your tin hat crooked today?
Oh, wow, gotta get me one o' them!