RE: computer pain
Hi Stathis/Jamie et al.
I've been busy else where in self-preservation mode deleting emails
madly .frustrating, with so many threads left hanging...oh well...but
I couldn't go past this particular dialog.
I am having trouble that you actually believe the below to be the case!
Lines of code that experience pain? Upon what law of physics is that
based?
Which one hurts more:
if (INPUT A) >= '1' then {
OUPUT "OUCH!"
}
or
if (INPUT A) >= '1' then {
OUPUT "OUCH!""OUCH!""OUCH!""OUCH!"
}
or
if (INPUT A) >= '10' then {
OUPUT "OUCH!""OUCH!""OUCH!""OUCH!"
}
Also: In a distributed applicationIf I put the program on earth, the
input on Mars and the CPU on the moon, which bit actually does the hurting
and when? It's still a program, still running - functionally the same
(time delays I know - not quite the same... but you get the idea).
The idea is predicated on the proven non-existance of a physical mechanism
for experience - that it somehow equates with manipulation of abstract
symbols as information rather than the fabric of reality as information. -
That pretending to be a neuron necessarily results in everything that a
neuron participates in as a chunk of matter.
It also completely ignores the ROLE of the experiences. There's a reason
for them. Unless you know the role you cannot assume that the software
model will inherit that role. With no role why bother with it? I don't
have to put "OUCH!""OUCH!""OUCH!" in the above.
What you are talking about is 'strong-AI' --- its functionalist
assumptions need to be carefully considered.
Another issue: If a life-like artefact visibly behaves like it is in agony
the only thing actually getting hurt are the humans watching it, who have
real experiences and empathy based on real qualia. It might be OK if it
were play. But otherwise? h.
cheers,
colin
>
>
> Jamie,
>
> I basically agree with your appraisal of the differences
> between living brains and digital computers. However, it
> should be possible for a general purpose computer to
> emulate the behaviour of a biological system in
> software. After all, biological systems are just
> comprised of matter following the laws of
> physics, which are well understood and deterministic
> at the size scales of interest.
> When it comes to neural tissue, the emulation should be
> able to replace the original provided that it is run on
> sufficiently fast hardware and has appropriate
> interfaces for input and output.
>
> While it would be extremely difficult to emulate a
> particular human brain (as in "mind uploading"), it should
> be easier to emulate a simplified generic brain, and easier
> again to emulate a single simplified perceptual function,
> such as pain.This means that it should be possible to store
> on a hard disk lines of code which, when
> run on a PC, will result in the program experiencing pain;
> perhaps excruciating pain beyond what
> humans can imagine, if certain parameters in the program
> are appropriately chosen. What might a simple example of
> such code look like? Should we try to determine what
> the painful programs are as a matter of urgency,
> in order to avoid using them in
> subroutines in other programs?
>
> Stathis Papaioannou
>
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To post to this group, send email to everything-list@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en
-~--~~~~--~~--~--~---
Re: Evil ?
1Z wrote: > > Brent Meeker wrote: >> 1Z wrote: >>> Stathis Papaioannou wrote: Bruno Marchal writes: > Le 12-déc.-06, à 11:16, Stathis Papaioannou a écrit : > >> Bruno Marchal writes (quoting Tom Caylor): >> In my view, your motivation is not large enough. I am also motivated by a problem: the problem of evil. I don't think the real problem of evil is solved or even really addressed with comp. This is because comp cannot define evil correctly. I will try to explain this more. >>> I agree that the problem of evil (and thus the equivalent problem of >>> Good) is interesting. Of course it is not well addressed by the two >>> current theories of everything: Loop gravity and String theory. With >>> that respect the comp hyp can at least shed some light on it, and of >>> course those "light" are of the platonic-plotinus type where the >>> notion >>> of goodness necessitates the notion of truth to begin with. I say more >>> below. >> Surely you have to aknowledge that there is a fundamental difference >> between matters of fact and matters of value. > Yes. Sure. And although I think that science is a value by itself, I am > not sure any scientific proposition can be used in judging those value. > But then, I also believe that this last sentence can be proved in comp > theories. > > > >> Science can tell us how to >> make a nuclear bomb and the effects a nuclear explosion will have on >> people >> and the environment, but whether it is "good" or "bad" to use such a >> weapon >> is not an empirical question at all. > Hmmm. This is not entirely true. We can test pain killer on people, > and we can see in scientific publication statements like "the drugs X > seem to provide help to patient suffering from disease Y". > Then it can be said that dropping a nuclear bomb on a city is bad for > such or such reason, and that it can be "good" in preventing bigger use > of nuclear weapon, etc. Again, we don't have too define good and bad > for reasoning about it once we agree on some primitive proposition > (that being rich and healthy is better than being poor and sick for > example). OK, but the point is that the basic definition of "bad" is arbitrary. >>> That isn't "just true" >>> It might seem that there would be some consensus, for example that torturing innocent people is an example of "bad", but it is possible to assert without fear of logical or empirical contradiction that torturing innocent people is good. >>> People don't want to be tortured. Isn't that empirical proof? >>> There are people in the world who do in fact think there is nothing wrong with torture and although they are not very nice peopel, they are not as a result of having such a belief deluded. >>> I think they are. Can you prove they are not? >>> > Recall that even the (although very familiar) notion of natural numbers > or integers cannot be defined unambiguously in science. Science asks us > only to be clear on primitive principles so that we can share some > reasoning on those undefinable entities. But there is a big difference between Pythagoras saying 17 is prime and Pythagoras saying that eating beans is bad. You can't say that "prime" and "bad" are equivalent in that they both need to be axiomatically defined. >>> Badness can be axiomatically defined (treating people as means rather >>> than ends, >>> acting on a maxim you would not wish to be universal law, not >>> doing as you would be done by, causaing unnecessary suffering). >> But such a definition doesn't make it so. >> >> I think discussions of good and evil go astray because they implicitly >> assume there is some objective good and evil. In fact all values are >> personal, only individuals experience suffering and joy. > > Only individuals can add numbers up, that doesn't make maths > subjective. That depends on how you mean "subjective". Math is objective in the sense that everybody agrees on it. But it's subjective in the sense that it depends on minds (subjects). Good and evil are not even objective in the sense of universal agreement, except possibly in the self-referential form such as, "My suffering is bad." So I think concepts of good and evil need to be built on the more fundamental personal vales. > >> Rules such as Kant's (which by the way says you shouldn't treat people >> *only* as ends) are attempts to derive social, ethical rules that provide >> for the realization of individual values. > > Kant's is explicitly more than that. Sure. I was just correcting the common misquote. > >> But individuals differ and so ethical rules always have exceptions in >> practice. > > All that means is that you can't have rules along the lines > of "don't tie anyon
Re: Evil ? (was: Hypostases
Brent Meeker wrote: > 1Z wrote: > > > > Stathis Papaioannou wrote: > >> Bruno Marchal writes: > >>> Le 12-déc.-06, à 11:16, Stathis Papaioannou a écrit : > >>> > > Bruno Marchal writes (quoting Tom Caylor): > > >> In my view, your motivation is not large enough. I am also motivated > >> by a problem: the problem of evil. I don't think the real problem of > >> evil is solved or even really addressed with comp. This is because > >> comp cannot define evil correctly. I will try to explain this more. > > > > I agree that the problem of evil (and thus the equivalent problem of > > Good) is interesting. Of course it is not well addressed by the two > > current theories of everything: Loop gravity and String theory. With > > that respect the comp hyp can at least shed some light on it, and of > > course those "light" are of the platonic-plotinus type where the > > notion > > of goodness necessitates the notion of truth to begin with. I say more > > below. > Surely you have to aknowledge that there is a fundamental difference > between matters of fact and matters of value. > >>> > >>> Yes. Sure. And although I think that science is a value by itself, I am > >>> not sure any scientific proposition can be used in judging those value. > >>> But then, I also believe that this last sentence can be proved in comp > >>> theories. > >>> > >>> > >>> > Science can tell us how to > make a nuclear bomb and the effects a nuclear explosion will have on > people > and the environment, but whether it is "good" or "bad" to use such a > weapon > is not an empirical question at all. > >>> > >>> Hmmm. This is not entirely true. We can test pain killer on people, > >>> and we can see in scientific publication statements like "the drugs X > >>> seem to provide help to patient suffering from disease Y". > >>> Then it can be said that dropping a nuclear bomb on a city is bad for > >>> such or such reason, and that it can be "good" in preventing bigger use > >>> of nuclear weapon, etc. Again, we don't have too define good and bad > >>> for reasoning about it once we agree on some primitive proposition > >>> (that being rich and healthy is better than being poor and sick for > >>> example). > >> OK, but the point is that the basic definition of "bad" is arbitrary. > > > > That isn't "just true" > > > >> It might seem > >> that there would be some consensus, for example that torturing innocent > >> people > >> is an example of "bad", but it is possible to assert without fear of > >> logical or > >> empirical contradiction that torturing innocent people is good. > > > > People don't want to be tortured. Isn't that empirical proof? > > > >> There are people > >> in the world who do in fact think there is nothing wrong with torture and > >> although > >> they are not very nice peopel, they are not as a result of having such a > >> belief deluded. > > > > I think they are. Can you prove they are not? > > > >>> Recall that even the (although very familiar) notion of natural numbers > >>> or integers cannot be defined unambiguously in science. Science asks us > >>> only to be clear on primitive principles so that we can share some > >>> reasoning on those undefinable entities. > >> But there is a big difference between Pythagoras saying 17 is prime and > >> Pythagoras > >> saying that eating beans is bad. You can't say that "prime" and "bad" are > >> equivalent > >> in that they both need to be axiomatically defined. > > > > Badness can be axiomatically defined (treating people as means rather > > than ends, > > acting on a maxim you would not wish to be universal law, not > > doing as you would be done by, causaing unnecessary suffering). > > But such a definition doesn't make it so. > > I think discussions of good and evil go astray because they implicitly assume > there is some objective good and evil. In fact all values are personal, only > individuals experience suffering and joy. Only individuals can add numbers up, that doesn't make maths subjective. > Rules such as Kant's (which by the way says you shouldn't treat people > *only* as ends) are attempts to derive social, ethical rules that provide for > the realization of individual values. Kant's is explicitly more than that. > But individuals differ and so ethical rules always have exceptions in > practice. All that means is that you can't have rules along the lines of "don't tie anyone up and spank them" since some people enjoy it. It doesn't stop you having more abstract rules. Like Kant's. > Everybody can agree that *their* suffering is bad; but that doesn't show > that making other people suffer is bad > - it is necessary for society to be able to punish people. "X is bad" doesn't mean you shouldn't do it under any circumstances. The alternative -- in this case letting criminals go unpunished -- might be worse. > Brent Meeker --~--~-
Re: Evil ? (was: Hypostases
1Z wrote: > > Stathis Papaioannou wrote: >> Bruno Marchal writes: >>> Le 12-déc.-06, à 11:16, Stathis Papaioannou a écrit : >>> Bruno Marchal writes (quoting Tom Caylor): >> In my view, your motivation is not large enough. I am also motivated >> by a problem: the problem of evil. I don't think the real problem of >> evil is solved or even really addressed with comp. This is because >> comp cannot define evil correctly. I will try to explain this more. > > I agree that the problem of evil (and thus the equivalent problem of > Good) is interesting. Of course it is not well addressed by the two > current theories of everything: Loop gravity and String theory. With > that respect the comp hyp can at least shed some light on it, and of > course those "light" are of the platonic-plotinus type where the > notion > of goodness necessitates the notion of truth to begin with. I say more > below. Surely you have to aknowledge that there is a fundamental difference between matters of fact and matters of value. >>> >>> Yes. Sure. And although I think that science is a value by itself, I am >>> not sure any scientific proposition can be used in judging those value. >>> But then, I also believe that this last sentence can be proved in comp >>> theories. >>> >>> >>> Science can tell us how to make a nuclear bomb and the effects a nuclear explosion will have on people and the environment, but whether it is "good" or "bad" to use such a weapon is not an empirical question at all. >>> >>> Hmmm. This is not entirely true. We can test pain killer on people, >>> and we can see in scientific publication statements like "the drugs X >>> seem to provide help to patient suffering from disease Y". >>> Then it can be said that dropping a nuclear bomb on a city is bad for >>> such or such reason, and that it can be "good" in preventing bigger use >>> of nuclear weapon, etc. Again, we don't have too define good and bad >>> for reasoning about it once we agree on some primitive proposition >>> (that being rich and healthy is better than being poor and sick for >>> example). >> OK, but the point is that the basic definition of "bad" is arbitrary. > > That isn't "just true" > >> It might seem >> that there would be some consensus, for example that torturing innocent >> people >> is an example of "bad", but it is possible to assert without fear of logical >> or >> empirical contradiction that torturing innocent people is good. > > People don't want to be tortured. Isn't that empirical proof? > >> There are people >> in the world who do in fact think there is nothing wrong with torture and >> although >> they are not very nice peopel, they are not as a result of having such a >> belief deluded. > > I think they are. Can you prove they are not? > >>> Recall that even the (although very familiar) notion of natural numbers >>> or integers cannot be defined unambiguously in science. Science asks us >>> only to be clear on primitive principles so that we can share some >>> reasoning on those undefinable entities. >> But there is a big difference between Pythagoras saying 17 is prime and >> Pythagoras >> saying that eating beans is bad. You can't say that "prime" and "bad" are >> equivalent >> in that they both need to be axiomatically defined. > > Badness can be axiomatically defined (treating people as means rather > than ends, > acting on a maxim you would not wish to be universal law, not > doing as you would be done by, causaing unnecessary suffering). But such a definition doesn't make it so. I think discussions of good and evil go astray because they implicitly assume there is some objective good and evil. In fact all values are personal, only individuals experience suffering and joy. Rules such as Kant's (which by the way says you shouldn't treat people *only* as ends) are attempts to derive social, ethical rules that provide for the realization of individual values. But individuals differ and so ethical rules always have exceptions in practice. Everybody can agree that *their* suffering is bad; but that doesn't show that making other people suffer is bad - it is necessary for society to be able to punish people. Brent Meeker --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to everything-list@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/everything-list?hl=en -~--~~~~--~~--~--~---
Re: computer pain
Stathis Papaioannou wrote:
>
> Brent Meeker writes:
>
>> I would say that many complex mechanical systems react to "pain" in a way
>> similar to simple animals. For example, aircraft have automatic shut downs
>> and fire extinguishers. They can change the flight controls to reduce
>> stress on structures. Whether they feel this "pain" is a different
>> question. I think they feel it if they incorporate it into a narrative to
>> which values are attached for purposes of learning ("Don't do that again, it
>> hurts."). But that's my theory of qualia - a speculative one.
>
> Pain mostly comes before learning. Infants are born with the
> ability to experience pain, so they learn to avoid activities which
> cause pain.
But the learning is a higher level thing. The experience has two levels. One
is just hardwired reactions, pulling your hand back from the fire. The
aircraft already has this, as to some very simple organisms. The other is part
of consciousness, which I speculate is creating a narrative in memory with
attached emotional values. Babies certainly feel pain in the first sense, but
they seem to have to learn to cry when hurt. I've accidentally stuck one of my
infant children when diapering them and gotten no reaction.
>It seems to be hardwired at a very basic level, which
> makes me think that it ought to be easier to implement in an AI than
> more complex cognitive processes and behaviours. But how would
> a behaviour such as an aircraft's reaction to a fire on board be
> characterised as "painful" in the way an infant putting its hand in a
> flame is painful? If the aircraft's experience is not painful, what can
> do to make it more like the baby's?
Add the narrative memory with values attached and then the ability to review
that memory when contemplating future actions.
Brent Meeker
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To post to this group, send email to everything-list@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en
-~--~~~~--~~--~--~---
Re: computer pain
Stathis, As I was reading your comments this morning, an example crossed my mind that might fit your description of in-place code lines that monitor 'disfunction' and exist in-situ as a 'pain' alert .. that would be error evaluating 'check-sum' computations. In a functional way, parallel check-summing, emulates at least the first part of an 'experience pain' rete .. the initial establishment of a signal message 'something is wrong'. What I'm getting at is that the question could be approached best by not retrofitting to 'experiential qualia' .. where we don't have a reasonable way to specify the different systems' 'experiencing', but we do have a way to identify analogs of process. For example - its possible to identify in architecture and materiales where 'points of highest stress' occur. The physicality of structures may indeed internally be "experiencing" higher-pressure nodes as 'pain' - where the only lack in the chain of our interaction with 'inanimate' structures, is OUR lack-of -wisdom in recognizing that those stress point are in fact 'pain-points' for those kinds of systems. For living systems, the nature of the neural connections is that the communication lines are still raw and open - back to the locus of the problem (pain site). In non-living structures, any break or disruption totally shuts down the back-reporting -- 'pain' disappears when all communication 'about' the pain-source is taken away or simply breaks down. Jamie --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to everything-list@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/everything-list?hl=en -~--~~~~--~~--~--~---
Re: Evil ? (was: Hypostases (was: Natural Order & Belief)
Bruno Marchal wrote: > > It might seem > > that there would be some consensus, for example that torturing > > innocent people > > is an example of "bad", but it is possible to assert without fear of > > logical or > > empirical contradiction that torturing innocent people is good. > > I disagree. Mainly for the reason alluded above. Please note I > understand that there is no purely logical contradiction (f) in > asserting that "torture" is good, but the purely logical operates at > the third person level, in which there is no "pain" at all. Once you > take incompleteness into account this should be much less evident, and > much more fuzzy. There is nothing illogical with an altimeter (in a > plane) giving a wrong information (like the plane is at altitude = > 1000, instead of the correct 500), but you can understand this can lead > to a catastrophe. Assuming catastrophes are bad. But that hardly show that falsehood and evil are identical, or even co-extensive. There can be good falsehood (comforting illusions) and ills that have nothing to do with falsehood. --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to everything-list@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/everything-list?hl=en -~--~~~~--~~--~--~---
Re: Hello all - My Theory of Everything
Le 13-déc.-06, à 02:45, Russell Standish a écrit : > Essentially that is the Occam razor theorem. Simpler universes have > higher probability. In the ASSA(*) realm I can give sense to this. I think Hal Finney and Wei Dai have defended something like this. But in the comp RSSA(**) realm, strictly speaking even the notion of "one" universe (even considered among other universes or in a multiverse à-la Deutsch) does not make sense unless the comp substitution level is *very* low. Stable appearances of local worlds emerge from *all* computations making all apparent (and thus sufficiently complex) world not "turing emulable". Recall that "I am a machine" entails "the apparent universe cannot be a machine" (= cannot be turing-emulable (cf UDA(***)). Bruno For the new people I recall the acronym: (*) ASSA = absolute self-sampling assumption (**) RSSA = relative self-sampling assumption The SSA idea is in the ASSA realm comes from Nick Bostrom, if I remember correctly. (***) UDA: see for example http://iridia.ulb.ac.be/~marchal/publications/SANE2004MARCHAL.htm http://iridia.ulb.ac.be/~marchal/ --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to everything-list@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/everything-list?hl=en -~--~~~~--~~--~--~---
Re: Evil ? (was: Hypostases (was: Natural Order & Belief)
Le 13-déc.-06, à 02:01, Stathis Papaioannou a écrit : > OK, but the point is that the basic definition of "bad" is arbitrary. Perhaps, but honestly I am not sure. In acomp, we can define a (very platonist) notion of "bad". The simpler and stronger one is just the falsity "f". Then Bf, BBf, BBBf, f, Bf, etc. gives a sequence of less and less badness, which translated in the Z (material) hypostases gives the Df, DDf, DDDf, f, Df ... which are better candidates for that notion of badness. (recall that G does not prove Bf -> f, and that G* proves DBf (the astonishing godelian consistency of being inconsistent). (note also that G* *does* prove Bf -> f). > It might seem > that there would be some consensus, for example that torturing > innocent people > is an example of "bad", but it is possible to assert without fear of > logical or > empirical contradiction that torturing innocent people is good. I disagree. Mainly for the reason alluded above. Please note I understand that there is no purely logical contradiction (f) in asserting that "torture" is good, but the purely logical operates at the third person level, in which there is no "pain" at all. Once you take incompleteness into account this should be much less evident, and much more fuzzy. There is nothing illogical with an altimeter (in a plane) giving a wrong information (like the plane is at altitude = 1000, instead of the correct 500), but you can understand this can lead to a catastrophe. Any BB...Bf can be seen as a promise for a catastrophe. > There are people > in the world who do in fact think there is nothing wrong with torture > and although > they are not very nice peopel, they are not as a result of having such > a belief deluded. Honestly I doubt it. Of course some people can believe that torture can be good for their own life, in case torture can prevent the enemy to drop some bomb. Of course some people are cynical and can, like Sade, defend torture with the (wrong imo) idea that nature "defends" the right of those who have the powers and thus that they have the right to follow their sexual perverse compulsion, but this could mean that they are inconsistent (they have some BBB...Bf as implicit belief). Then from the divine (starred G*) pov, they are (globally) inconsistent (although cannot know it). > >> Recall that even the (although very familiar) notion of natural >> numbers >> or integers cannot be defined unambiguously in science. Science asks >> us >> only to be clear on primitive principles so that we can share some >> reasoning on those undefinable entities. > > But there is a big difference between Pythagoras saying 17 is prime > and Pythagoras > saying that eating beans is bad. You can't say that "prime" and "bad" > are equivalent > in that they both need to be axiomatically defined. Hmmm... "prime" and "bad" cannot be equivalent in that sense. But "being a natural number" and "bad" can. The nuance is that I grant the notion of natural number before defining "prime". But my belief in natural numbers (my belief in the standard model of Peano Arithmetic, Arithmetical truth) is as hard, even impossible, to define than is the notion of truth, good, etc. Defining "Prime" is easy: (~(x = 1) & Ay(y divides x) -> (y = 1 v y = x)) where (a divides b) is a macro for Ez(az = b). Defining "number" is just not possible actually. Even with a richer theory or second order logic you will have to rely implicitly on the standard model of the higher theory, which is less palatable than the "standard model" of PA. > The problem is that some people think "good" and "bad" are on a par > with > descriptive terms that every sentient species, regardless of their > psychology, > could agree on. They are not. Not in any normative sense. But once we bet on a theory (like comp), then we get mathematical tools which can provide general explanation of what is bad, and also explain why such definition cannot be normative, making the bad/good distinctions an ideal goal for complex sufficiently self-sustaining machines societies. > Every sentient species would agree that a > nuclear bomb going off in your face will kill you, Bad example for this list! (CF quantum immortality or comp immortality!). But ok, this is besides the point. > but some would say this was > good and others would say it was bad. Yes, but unless people are insane, most will give or try to give a ratio. In such case it is a question of utility with respect of some notion of good and bad. It is not related with the hardness to define completely what is good and what is bad. Like killing. Killing can be considered as bad but can be accepted in self-defense. > I think a message spelt out across the sky by stars simultaneously > going > nova would probably do it for me. I would bet I'm dreaming instead ... :) > I would at least believe that these were > beings with godlike powers responsible,
Re: Evil ? (was: Hypostases (was: Natural Order & Belief)
Stathis Papaioannou wrote: > Bruno Marchal writes: > > > > Le 12-déc.-06, à 11:16, Stathis Papaioannou a écrit : > > > > > > > > > > > Bruno Marchal writes (quoting Tom Caylor): > > > > > >>> In my view, your motivation is not large enough. I am also motivated > > >>> by a problem: the problem of evil. I don't think the real problem of > > >>> evil is solved or even really addressed with comp. This is because > > >>> comp cannot define evil correctly. I will try to explain this more. > > >> > > >> > > >> I agree that the problem of evil (and thus the equivalent problem of > > >> Good) is interesting. Of course it is not well addressed by the two > > >> current theories of everything: Loop gravity and String theory. With > > >> that respect the comp hyp can at least shed some light on it, and of > > >> course those "light" are of the platonic-plotinus type where the > > >> notion > > >> of goodness necessitates the notion of truth to begin with. I say more > > >> below. > > > > > > Surely you have to aknowledge that there is a fundamental difference > > > between matters of fact and matters of value. > > > > > > Yes. Sure. And although I think that science is a value by itself, I am > > not sure any scientific proposition can be used in judging those value. > > But then, I also believe that this last sentence can be proved in comp > > theories. > > > > > > > > > Science can tell us how to > > > make a nuclear bomb and the effects a nuclear explosion will have on > > > people > > > and the environment, but whether it is "good" or "bad" to use such a > > > weapon > > > is not an empirical question at all. > > > > > > Hmmm. This is not entirely true. We can test pain killer on people, > > and we can see in scientific publication statements like "the drugs X > > seem to provide help to patient suffering from disease Y". > > Then it can be said that dropping a nuclear bomb on a city is bad for > > such or such reason, and that it can be "good" in preventing bigger use > > of nuclear weapon, etc. Again, we don't have too define good and bad > > for reasoning about it once we agree on some primitive proposition > > (that being rich and healthy is better than being poor and sick for > > example). > > OK, but the point is that the basic definition of "bad" is arbitrary. That isn't "just true" > It might seem > that there would be some consensus, for example that torturing innocent people > is an example of "bad", but it is possible to assert without fear of logical > or > empirical contradiction that torturing innocent people is good. People don't want to be tortured. Isn't that empirical proof? > There are people > in the world who do in fact think there is nothing wrong with torture and > although > they are not very nice peopel, they are not as a result of having such a > belief deluded. I think they are. Can you prove they are not? > > Recall that even the (although very familiar) notion of natural numbers > > or integers cannot be defined unambiguously in science. Science asks us > > only to be clear on primitive principles so that we can share some > > reasoning on those undefinable entities. > > But there is a big difference between Pythagoras saying 17 is prime and > Pythagoras > saying that eating beans is bad. You can't say that "prime" and "bad" are > equivalent > in that they both need to be axiomatically defined. Badness can be axiomatically defined (treating people as means rather than ends, acting on a maxim you would not wish to be universal law, not doing as you would be done by, causaing unnecessary suffering). > > > You could say that "I believe blowing people up is bad" is a statement > > > of > > > empirical fact, either true or false depending on whether you are > > > accurately > > > reporting your belief. However, "blowing people up is bad" is a > > > completely > > > different kind of statement which no amount of empirical evidence has > > > any > > > bearing on. > > > > > > > > It really depends on the axioms of your theory. A theory of good and > > bad for a lobian machine can be based on the idea of 3-surviving or > > 1-surviving, etc. And then we can reason. > > Now I do agree with you that good and bad can probably not be defined > > intrinsically in a mathematical way. But a richer lobian machine can > > define some notion of self-referential correctness for a less rich > > lobian machine and then reason about it, and then lift the result in > > some interrogative way about herself. > > Some suicide phenomenon with animals could be explained in such a way. > > You have the Parfit book "reason and persons". There are many pieces of > > valid reasoning (and non normative) on ethical points in that book. > > Science can handle values and relation between values as far as it does > > not judge normatively those values. > > > > > If you survey a million people and all of them believe that "blowing > > > up people is bad", you have shown that "most people believe that
RE: computer pain
Brent Meeker writes:
> I would say that many complex mechanical systems react to "pain" in a way
> similar to simple animals. For example, aircraft have automatic shut downs
> and fire extinguishers. They can change the flight controls to reduce stress
> on structures. Whether they feel this "pain" is a different question. I
> think they feel it if they incorporate it into a narrative to which values
> are attached for purposes of learning ("Don't do that again, it hurts.").
> But that's my theory of qualia - a speculative one.
Pain mostly comes before learning. Infants are born with the
ability to experience pain, so they learn to avoid activities which
cause pain. It seems to be hardwired at a very basic level, which
makes me think that it ought to be easier to implement in an AI than
more complex cognitive processes and behaviours. But how would
a behaviour such as an aircraft's reaction to a fire on board be
characterised as "painful" in the way an infant putting its hand in a
flame is painful? If the aircraft's experience is not painful, what can
do to make it more like the baby's?
Stathis Papaioannou
_
Be one of the first to try Windows Live Mail.
http://ideas.live.com/programpage.aspx?versionId=5d21c51a-b161-4314-9b0e-4911fb2b2e6d
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To post to this group, send email to everything-list@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en
-~--~~~~--~~--~--~---
RE: computer pain
Jamie, I basically agree with your appraisal of the differences between living brains and digital computers. However, it should be possible for a general purpose computer to emulate the behaviour of a biological system in software. After all, biological systems are just comprised of matter following the laws of physics, which are well understood and deterministic at the size scales of interest. When it comes to neural tissue, the emulation should be able to replace the original provided that it is run on sufficiently fast hardware and has appropriate interfaces for input and output. While it would be extremely difficult to emulate a particular human brain (as in "mind uploading"), it should be easier to emulate a simplified generic brain, and easier again to emulate a single simplified perceptual function, such as pain. This means that it should be possible to store on a hard disk lines of code which, when run on a PC, will result in the program experiencing pain; perhaps excruciating pain beyond what humans can imagine, if certain parameters in the program are appropriately chosen. What might a simple example of such code look like? Should we try to determine what the painful programs are as a matter of urgency, in order to avoid using them in subroutines in other programs? Stathis Papaioannou > Date: Tue, 12 Dec 2006 23:19:05 -0800 > From: [EMAIL PROTECTED] > To: everything-list@googlegroups.com > Subject: Re: computer pain > > > Stathis, > > The reason for lack of responses is that your idea > goes directly to illuminating why AI systems - as > promoulgated under current designs of software > running in hardware matrices - CANNOT emulate living > systems. It an issue that AI advocates intuitively > and scrupulously AVOID. > > "Pain" in living systems isn't just a self-sensor > of proper/improper code functioning, it is an embedded > registration of viable/disrupted matrix state. > > And that is something that no current human contrived > system monitors as a CONCURRENT property of software. > > For example, we might say that central processors > regularly 'display pain' .. that we designers/users > recognize as excess heat .. that burn out mother boards. > The equipment 'runs a high fever', in other words. > > But where living systems are multiple functioning systems > and have internal ways of guaging and reacting locally and > biochemically vis a vis both to the variance and retaining > sufficient good-operations while bleeding off 'fever', > "hardware" systems have no capacity to morph or adapt > itself structurally and so keep on burning up or wait > for external aware-structures to command them to stop > operating for a while and let the equipment cool down. > > I maintain that living systems are significantly designed where > hardware IS software, and so have a capacity for local > adaptive self-sensitivity, that human 'contrived' HW/SW systems > don't and mostly .. can't. > > Jamie Rose > > > Stathis Papaioannou wrote: > > > > No responses yet to this question. It seems to me a straightforward > > consequence of computationalism that we should be able to write a program > > which, when run, will experience pain, and I suspect that this would be a > > substantially simpler program than one demonstrating general intelligence. > > It > > would be very easy to program a computer or build a robot that would behave > > just like a living organism in pain, but I'm not sure that this is nearly > > enough to > > ensure that it is in fact experiencing pain. Any ideas, or references to > > sources > > that have considered the problem? > > > > Stathis Papaioannou > > > > _ Be one of the first to try Windows Live Mail. http://ideas.live.com/programpage.aspx?versionId=5d21c51a-b161-4314-9b0e-4911fb2b2e6d --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to everything-list@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/everything-list?hl=en -~--~~~~--~~--~--~---
Re: Hello all - My Theory of Everything
On Wed, Dec 13, 2006 at 09:14:36AM -, William wrote: > > I think I'm following your reasoning here, this theorem could also be > used to prove that any probability distribution for universes, which > gives a lower or equal probability to a system with fewer information; > must be wrong. Right ? Essentially that is the Occam razor theorem. Simpler universes have higher probability. > > But in this case, could one not argue that there is only a small number > (out of the total) of "higher" universes containing an SAS, and then > rephrase the statement to "we are not being simulated by another SAS" ? > By "higher" I gather you mean more complex. But I think you are implicitly assuming that a more complex universe is needed to simulate this one, which I think is wrong. All that is needed is Turing completeness, which even very simple universes have (for instance Conway's Game of Life). Cheers PS - I'm off tomorrow for the annual family pilgrimage, so I'll be rather quiet on this list for the next month. A/Prof Russell Standish Phone 0425 253119 (mobile) Mathematics UNSW SYDNEY 2052 [EMAIL PROTECTED] Australiahttp://www.hpcoders.com.au --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to everything-list@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/everything-list?hl=en -~--~~~~--~~--~--~---
Re: Hello all - My Theory of Everything
Russell Standish schreef: > On Mon, Dec 11, 2006 at 03:26:59PM -0800, William wrote: > > > > > If the universe is computationallu simulable, then any universal > > > Turing machine will do for a "higher hand". In which case, the > > > information needed is simply the shortest possible program for > > > simulating the universe, the length of which by definition is the > > > information content of the universe. > > > > What I meant to compare is 2 situations (I've taken an SAS doing the > > simulations for now although i do not think it is required): > > > > 1) just our universe A consisting of minimal information > > 2) An interested SAS in another universe wants to simulate some > > universes; amongst which is also universe A, ours. > > > > Now we live in universe A; but the question we can ask ourselves is if > > we live in 1) or 2). (Although one can argue there is no actual > > difference). > > > > Nevertheless, my proposition is that we live in 1; since 2 does exist > > but is less probable than 1. > > > > information in 1 = inf(A) > > information in 2 = inf(simulation_A) + inf(SAS) + inf(possible other > > stuff) = inf(A) + inf(SAS) + inf(possible other stuff) > inf(A) > > > > You're still missing the point. If you sum over all SASes and other > computing devices capable of simulating universe A, the probability of > being in a simulation of A is identical to simply being in universe A. > > This is actually a theorem of information theory, believe it or not! I think I'm following your reasoning here, this theorem could also be used to prove that any probability distribution for universes, which gives a lower or equal probability to a system with fewer information; must be wrong. Right ? But in this case, could one not argue that there is only a small number (out of the total) of "higher" universes containing an SAS, and then rephrase the statement to "we are not being simulated by another SAS" ? --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Everything List" group. To post to this group, send email to everything-list@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/everything-list?hl=en -~--~~~~--~~--~--~---
Re: computer pain
James N Rose wrote:
> Stathis,
>
> The reason for lack of responses is that your idea
> goes directly to illuminating why AI systems - as
> promoulgated under current designs of software
> running in hardware matrices - CANNOT emulate living
> systems. It an issue that AI advocates intuitively
> and scrupulously AVOID.
>
> "Pain" in living systems isn't just a self-sensor
> of proper/improper code functioning, it is an embedded
> registration of viable/disrupted matrix state.
>
> And that is something that no current human contrived
> system monitors as a CONCURRENT property of software.
>
> For example, we might say that central processors
> regularly 'display pain' .. that we designers/users
> recognize as excess heat .. that burn out mother boards.
> The equipment 'runs a high fever', in other words.
>
> But where living systems are multiple functioning systems
> and have internal ways of guaging and reacting locally and
> biochemically vis a vis both to the variance and retaining
> sufficient good-operations while bleeding off 'fever',
> "hardware" systems have no capacity to morph or adapt
> itself structurally and so keep on burning up or wait
> for external aware-structures to command them to stop
> operating for a while and let the equipment cool down.
>
> I maintain that living systems are significantly designed where
> hardware IS software, and so have a capacity for local
> adaptive self-sensitivity, that human 'contrived' HW/SW systems
> don't and mostly .. can't.
>
> Jamie Rose
>
>
> Stathis Papaioannou wrote:
>> No responses yet to this question. It seems to me a straightforward
>> consequence of computationalism that we should be able to write a program
>> which, when run, will experience pain, and I suspect that this would be a
>> substantially simpler program than one demonstrating general intelligence. It
>> would be very easy to program a computer or build a robot that would behave
>> just like a living organism in pain, but I'm not sure that this is nearly
>> enough to
>> ensure that it is in fact experiencing pain. Any ideas, or references to
>> sources
>> that have considered the problem?
>>
>> Stathis Papaioannou
I would say that many complex mechanical systems react to "pain" in a way
similar to simple animals. For example, aircraft have automatic shut downs and
fire extinguishers. They can change the flight controls to reduce stress on
structures. Whether they feel this "pain" is a different question. I think
they feel it if they incorporate it into a narrative to which values are
attached for purposes of learning ("Don't do that again, it hurts."). But
that's my theory of qualia - a speculative one.
Brent Meeker
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To post to this group, send email to everything-list@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en
-~--~~~~--~~--~--~---
