Can a computer be "conscious"?

[John Cuthber]

Isn't the answer to this question "Yes, I have one between my ears"?

I think we need a definition of computer as well as consious.

If my brain (and as much of the nervous/ sensory system as is required) can be mapped onto a turing machine then a computer can be consious (at least as much as I can).

If not then it can't (at least for the definition of consious that says it means "acting like a brain").

 

Is the brain a turing machine?

If not, what does it do that a turing machine can't?

(and I know that one possible answer to that is that a brain can be consious- but that's begging the question).

[immortal]

 

Is the brain a turing machine?

If not, what does it do that a turing machine can't?

(and I know that one possible answer to that is that a brain can be consious- but that's begging the question).

 

The brain is not a Turing machine. The brain can say that a statement is true which a turing machine can't. This is the simplest example that I found on the net " The sum of two even numbers is always an even number". We know that this statement is obviously true but the program on the turing machine runs without halting. This is why Roger Penrose argues that the states of consciousness have an element of non-computability.

[John Cuthber]

The brain can say if some statements are true but there are some about which we are not sure (not least, the quesstion that forms the title of this thread).

I could write a program that, if asked "Is the sum of 2 even numbers an even number?" would reply "Yes".

It may well be the the fact that we con often answer general questions like this means that we are not turing machines- but it could be that we are somehow programed to recognise some questions and give the right answer.

[Vesna]

Interesting. The question of simulating or anyhow else producing consciousness ultimately comes down to determinism.

 

The brain is not a Turing machine. The brain can say that a statement is true which a turing machine can't. This is the simplest example that I found on the net " The sum of two even numbers is always an even number". We know that this statement is obviously true but the program on the turing machine runs without halting. This is why Roger Penrose argues that the states of consciousness have an element of non-computability.

 

quantum or turing machine? i'm with you two in this one. i say not only is a brain quantum computer, but the whole universe is a quantum computer. it works on non-computability, just like Improbability Drive that powers the star ship of Zaphod Beeblebrox.

 

Douglas Adams: "There is a theory which states that if ever anyone discovers exactly what the Universe is for and why it is here, it will instantly disappear and be replaced by something even more bizarre and inexplicable.There is another theory which states that this has already happened."

 

 

immortal, i do not see how that works, could you present the logic behind that conclusion? i don't think "We know that this statement is obviously true". we can only know it to a certain degree by trying to calculate and check it out for real, which is what you assume program will do to infinity.

 

1.) we can advise computer to limit computation to 'certain degree' of precision, so it can answer in percentages of certainty rather than in absolutes.

 

2.) if we have any other way of "knowing for sure", then we can teach computer exactly that and AI could use same theorems and logic instead of blind calculation to answer the question. if there is a mathematical or logical proof for that, then theoretically AI should be able to calculate it as much as we could and be certain about it as much as we are.

 

 

I could write a program that, if asked "Is the sum of 2 even numbers an even number?" would reply "Yes".

It may well be the the fact that we con often answer general questions like this means that we are not turing machines- but it could be that we are somehow programed to recognise some questions and give the right answer.

 

you speak about AI in the context of someone's opinion, in order to trick some specific person at specific time about some supposed AI. i think it should be made clear what is the intention here - to make a really smart AI that can fool us all, or to make even the most stupid AI, but working on the same principles as living system?

 

do we want to trick ourselves or to discover the principles behind it?

Edited February 3, 2009 by Vesna

[bascule]

The brain is not a Turing machine. The brain can say that a statement is true which a turing machine can't.

 

If a Turing machine were to be conscious its thought process would be bound no more by formal logic than ours is. Consider the case of a brain simulation.

 

This is why Roger Penrose argues that the states of consciousness have an element of non-computability.

 

Penrose, particularly in things like his Godelian argument, assumes the thought process of a "robot mathematician" is bound by formal logic because the underlying system is one based on formal logic.

 

Humans don't reason with formal logic (by default, of course we can learn it), and Penrose uses this to try to conclude that there's something "special" about us (i.e. thought is intrinsically tied to quantum waveform collapse)

 

However, Penrose's Godelian argument is not only flawed for other reasons but it makes little sense to treat consciousness itself as a formal logic system.

[immortal]

 

immortal, i do not see how that works, could you present the logic behind that conclusion? i don't think "We know that this statement is obviously true". we can only know it to a certain degree by trying to calculate and check it out for real, which is what you assume program will do to infinity.

 

1.) we can advise computer to limit computation to 'certain degree' of precision, so it can answer in percentages of certainty rather than in absolutes.

 

2.) if we have any other way of "knowing for sure", then we can teach computer exactly that and AI could use same theorems and logic instead of blind calculation to answer the question. if there is a mathematical or logical proof for that, then theoretically AI should be able to calculate it as much as we could and be certain about it as much as we are.

 

 

 

 

you speak about AI in the context of someone's opinion, in order to trick some specific person at specific time about some supposed AI. i think it should be made clear what is the intention here - to make a really smart AI that can fool us all, or to make even the most stupid AI, but working on the same principles as living system?

 

do we want to trick ourselves or to discover the principles behind it?

 

First of all let me say that we don't have a good definition for intelligence on which we all agree upon. So I think intelligence is something beyond space and time and it not physical and it is not an idea. As you can see I am a Platonist. I will clear it for you.

 

1. Well in mathematics you should'nt have doubts. You can not say that the statement is 50% true and 50% false. With this kind of a computer you can not prove anything. Mathematical statements are absolute.

 

2. I think the way humans know that it is true is because we use a non-algorithmic process to access those absolute truths. This is why Roger Penrose thinks that we need new physics. Once we have found out that non-algorithmic process then we can think of building a AI machine if possible.

---

Merged post follows:

Consecutive posts merged

 

immortal, i do not see how that works, could you present the logic behind that conclusion? i don't think "We know that this statement is obviously true". we can only know it to a certain degree by trying to calculate and check it out for real, which is what you assume program will do to infinity.

 

1.) we can advise computer to limit computation to 'certain degree' of precision, so it can answer in percentages of certainty rather than in absolutes.

 

2.) if we have any other way of "knowing for sure", then we can teach computer exactly that and AI could use same theorems and logic instead of blind calculation to answer the question. if there is a mathematical or logical proof for that, then theoretically AI should be able to calculate it as much as we could and be certain about it as much as we are.

 

 

 

 

you speak about AI in the context of someone's opinion, in order to trick some specific person at specific time about some supposed AI. i think it should be made clear what is the intention here - to make a really smart AI that can fool us all, or to make even the most stupid AI, but working on the same principles as living system?

 

do we want to trick ourselves or to discover the principles behind it?

 

First of all let me say that we don't have a good definition for intelligence on which we all agree upon. So I think intelligence is something beyond space and time and it not physical and it is not an idea. As you can see I am a Platonist. I will clear it for you.

 

1. Well in mathematics you should'nt have doubts. You can not say that the statement is 50% true and 50% false. With this kind of a computer you can not prove anything. Mathematical statements are absolute.

 

2. I think the way humans know that it is true is because we use a non-algorithmic process to acsess those absolute truths. This is why Roger Penrose thinks that we need new physics. Once we have found out that non-algorithmic process then we can think of building a AI machine if possible.

[Vesna]

2. I think the way humans know that it is true is because we use a non-algorithmic process to acsess those absolute truths. This is why Roger Penrose thinks that we need new physics. Once we have found out that non-algorithmic process then we can think of building a AI machine if possible.

 

i agree with Roger, but math or logic is algorithmic process, very much. humans can only "know it" according to something, some theorem, some logical, mathematical or geometrical relation. what i'm trying to tell you is that as soon as you find me a theorem according to which humans are certain about it, i will show you how can i teach AI to think exactly like that.

 

computers are not just 'plus' and 'minus', but also 'xor' and 'or', computers can use the same logic and mathematics as we can, the question is how to teach them. if you teach them to be stupid, then they might compute it to infinity, but if you teach the same knowledge as you know, then they will know the same and be able to compute the same.

 

 

logic and mathematics are functions of efficiency, it is one thing computers can learn for certain, while emotions and feelings like creativity and melancholy is what we do not understand, description is subjective and a matter of opinion, which makes it vague as opposed to math and logic that are exact - deterministic - algorithmic.