In the far future, can a robot ever mimic all human behavior perfectly?
- Thread starter a_ht
- Start date
- Status
Because if a robot can attain the kind of cognitive abilities and thought processes that a person can, it will also have access to the super computing capabilities of a computer. It will be, in my mind, much like a human mind linked with a computer. This means it will have perfect recollection of events it deems important enough. A built in calculator that enables it to be more efficient than the smartest human when doing mathematical equations,.
-AntonK
-AntonK
No, because a robot, like humans, isn't ever going to made perfectly. Since perfection is never found, except in nature, then a robot will make errors as does all humankind.
Anytime free will and concious is mentioned the issue of ethics and morality becomes apparent. Will the a.i have to waste precious computing power making moral decisions? Or will it see sympathy as an illogical and irreverent flaw of humanity?
Because you're thinking of a.i acting human and on free will....that won't likely be the case. Acting human isn't neccessairly the second step in a.i progression, infact having free will and an active "concious" serves no real purpose in a.i.
As with anything costly, advanced a.i would be designed with total servitude in mind, not an experiment in digital life immitating an analog one.
Because you're thinking of a.i acting human and on free will....that won't likely be the case. Acting human isn't neccessairly the second step in a.i progression, infact having free will and an active "concious" serves no real purpose in a.i.
As with anything costly, advanced a.i would be designed with total servitude in mind, not an experiment in digital life immitating an analog one.
eburacum45
Valued Senior Member
I don't think that any AI created to be a slave could be truly intelligent or even truly conscious;
but I think there will be many different types of artificial mind, and they will each develop in new and unpredictable ways.
So, for a laugh, I will try to predict them.
You could have 'chinese room' minds; computers with a vast repertoire of human responses which it can select at a extremely rapid rate, so fast and convincing that most humans will accept this kind of machine as effectively conscious.
Other minds will not mimic humans, but be designed to learn new fact and process information spontaneously, while obeying human commands; these are the sort of robots Asimov imagined in his books. Being programmed to obey these machines will not spontaneously develop goals, or desires, or other emotions; they could be fantastically competent and have access to vast databases of information, but they will be less than human.
Other artificial minds may develop from attempts to emulate simple animal behaviour and learning; an AI might be organised like an ant hill, and ant hills are perfectly capable of developing their own goals over time; but they probably not conscious. If an anthill mind is given enough processing power it might develop an analogue to consciousness; but it might be radically diffent to any human consciousness.
Similarly artificial analogues of lizard, bird, mouse minds could be developed; eventually human minds could be emulated, which is the situation I think a_ht was anticipating in the original post.
These expected human emulations might be no better than our own, but they have the potential for upgrades; add a bigger calculator function, a bigger memory; some mouse and antbrain peripherals, a few more operating system languages, some more optical and auditory sense systems;
connect a hundred of these expanded minds together, and give them the plans of their own construction to work on...
artificial minds might exceed human minds in this respect more than any other- the ability to inspect and record the internal workings of its own mind, and the ability to redesign itself, could take the various types of artificial mind far beyond any human consciousness.
This sort of development may be so subtle and unexpected that we might not even recognise it when it happens.
-----------------
www.orionsarm.com
but I think there will be many different types of artificial mind, and they will each develop in new and unpredictable ways.
So, for a laugh, I will try to predict them.
You could have 'chinese room' minds; computers with a vast repertoire of human responses which it can select at a extremely rapid rate, so fast and convincing that most humans will accept this kind of machine as effectively conscious.
Other minds will not mimic humans, but be designed to learn new fact and process information spontaneously, while obeying human commands; these are the sort of robots Asimov imagined in his books. Being programmed to obey these machines will not spontaneously develop goals, or desires, or other emotions; they could be fantastically competent and have access to vast databases of information, but they will be less than human.
Other artificial minds may develop from attempts to emulate simple animal behaviour and learning; an AI might be organised like an ant hill, and ant hills are perfectly capable of developing their own goals over time; but they probably not conscious. If an anthill mind is given enough processing power it might develop an analogue to consciousness; but it might be radically diffent to any human consciousness.
Similarly artificial analogues of lizard, bird, mouse minds could be developed; eventually human minds could be emulated, which is the situation I think a_ht was anticipating in the original post.
These expected human emulations might be no better than our own, but they have the potential for upgrades; add a bigger calculator function, a bigger memory; some mouse and antbrain peripherals, a few more operating system languages, some more optical and auditory sense systems;
connect a hundred of these expanded minds together, and give them the plans of their own construction to work on...
artificial minds might exceed human minds in this respect more than any other- the ability to inspect and record the internal workings of its own mind, and the ability to redesign itself, could take the various types of artificial mind far beyond any human consciousness.
This sort of development may be so subtle and unexpected that we might not even recognise it when it happens.
-----------------
www.orionsarm.com
Last edited:
kriminal99
Registered Senior Member
AntonK said:
Humans already have a built in calculator.
The question is can a robot be built without the same limitations on consiousness that human beings have. For example could it consiously process 50 things at once as opposed to the like 7 things that a person can. Or instead perhaps its subconsiously driven impulses would be so precise that it would allow it to "intuit" the answer to 3453634634 * 223432 within seconds.
This is all assuming its designed anything like a human consiousness, which we don't know yet the specifics of.
a_ht said:
Human beings are no more than machines themselves. Its already clear to those who wish to know the specifics of how a human consiousness operates. So to answer your question, yes.
loophole said:
Interesting question....what is perfect human behavior on which A.I. can base a template to follow?
Facial
Valued Senior Member
kriminal99's right.
Once we know how the individual neurons in our brain interact, then we can mimic that behavior with transistors, or perhaps something else. The human brain is almost infinite in complexity, but not truly infinite. So it is within our grasp of understanding.
The most powerful supercomputers today can already fully emulate a spider brain.
Once we know how the individual neurons in our brain interact, then we can mimic that behavior with transistors, or perhaps something else. The human brain is almost infinite in complexity, but not truly infinite. So it is within our grasp of understanding.
The most powerful supercomputers today can already fully emulate a spider brain.
Cosmic,
The question was whether it could mimic human behaviour perfectly not that it was perfect. This implies that it should be imperfect, just like humans. Given that the human brain is primarily a collection of bio-electrical connections then it seems highly likely that a machine could be built to mimic the same combinations and interactions.
Cris
The question was whether it could mimic human behaviour perfectly not that it was perfect. This implies that it should be imperfect, just like humans. Given that the human brain is primarily a collection of bio-electrical connections then it seems highly likely that a machine could be built to mimic the same combinations and interactions.
Cris
cosmictraveler said:
So, what made humans?
Nature isn't perfect, if it were, it would have created the perfect balance of perfect life forms, eons ago (which it probably did). Come to think of it, even humans aren't all that bad... I mean, it's only during the past 100 years or so ago that we started destroying the planet. But anyway, it just goes to show that everybody and everything makes mistakes and those super savvy robots of future come, will certainly re-inforce the proof of that.
Facial said:
I would say, 'open ended'--always able to learn.
Facial.
What does almost infinite mean? Infinity is simply something that has no boundary. The human brain has very finite limits with some 200 Billion neurons and several trillion synapses (connections). We already exceed those numbers with the number of transistors on a set of computer IC chips. But it isn’t the number of components that is important but the processing speed.
Given that each neuron fires at around 300 Hz then 200 billion neurons is the equivalent of 20,000 Intel 3GHz Pentium IV processors operating in parallel. That’s a rather large number of processors to link up and make work efficiently, although not entirely undoable today. However, Moore’s law has held well since the 1940’s and demonstrates that CPU power doubles every 12 to 18 months. In this case this 20,000 value comes down to 1000 by 2010 and perhaps just a single chip by 2025.
In essence we will highly likely have the convenient raw processing power of the human brain within the next decade, even if it takes longer it is very likely in the lifetime for most of us. As we approach this type of power then we can more realistically experiment with the massive types of real-time processing that the brain performs in normal operation. The issues then become one of software engineering and the generation of the algorithms that duplicate those in the human brain, and I can’t give any meaningful estimate for that work.
So I’m pretty confident that we will be able to duplicate the human brain in silicon and software relatively soon but the more interesting issue is about what comes next. Assuming that technology continues to double CPU power every 18 months then even when we have achieved human bran power then the following year we will have exceeded the human brain by 100%. It seems likely that before this century is over humans will be by far only a minor intelligence on the planet.
The other interesting issue will be to observe when self-awareness kicks in or will it be possible to create algorithms of pure intelligence without self-awareness – I don’t think that the former can exist without the latter being inevitable.
Cris
What does almost infinite mean? Infinity is simply something that has no boundary. The human brain has very finite limits with some 200 Billion neurons and several trillion synapses (connections). We already exceed those numbers with the number of transistors on a set of computer IC chips. But it isn’t the number of components that is important but the processing speed.
Given that each neuron fires at around 300 Hz then 200 billion neurons is the equivalent of 20,000 Intel 3GHz Pentium IV processors operating in parallel. That’s a rather large number of processors to link up and make work efficiently, although not entirely undoable today. However, Moore’s law has held well since the 1940’s and demonstrates that CPU power doubles every 12 to 18 months. In this case this 20,000 value comes down to 1000 by 2010 and perhaps just a single chip by 2025.
In essence we will highly likely have the convenient raw processing power of the human brain within the next decade, even if it takes longer it is very likely in the lifetime for most of us. As we approach this type of power then we can more realistically experiment with the massive types of real-time processing that the brain performs in normal operation. The issues then become one of software engineering and the generation of the algorithms that duplicate those in the human brain, and I can’t give any meaningful estimate for that work.
So I’m pretty confident that we will be able to duplicate the human brain in silicon and software relatively soon but the more interesting issue is about what comes next. Assuming that technology continues to double CPU power every 18 months then even when we have achieved human bran power then the following year we will have exceeded the human brain by 100%. It seems likely that before this century is over humans will be by far only a minor intelligence on the planet.
The other interesting issue will be to observe when self-awareness kicks in or will it be possible to create algorithms of pure intelligence without self-awareness – I don’t think that the former can exist without the latter being inevitable.
Cris
Cris said:
And that, is really the most frightening issue.
Cris said:
I agree completely, but the question was "the far future".
Now that I think of it though, I doubt it will take more than a few generations before people forget what it was like not to be "enhanced" or whatever. Well, that is if there isn't a millenia or few of culture wars with cyborgs and fleshy fundies battling it out.
- Status