Total Posts:86|Showing Posts:1-30|Last Page
Jump to topic:

Conscious AI

Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
10/29/2016 6:28:35 PM
Posted: 1 month ago
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?
Genius_Intellect
Posts: 339
Add as Friend
Challenge to a Debate
Send a Message
10/29/2016 10:02:46 PM
Posted: 1 month ago
"Consciousness" is just a naturalistic way of saying "soul". The soul is an illusion, created when the different parts of the brain work together properly. Descartes said, "I think, therefore I am," but he had it backwards. It should be: "I am, because I think." There's no magic spark inside of us, just a brain being animated by electrical signals and hormones. AI will be no different.
keithprosser
Posts: 1,964
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 9:48:29 AM
Posted: 1 month ago
Unfortunately no one has the faintest idea how to make a computer or machine conscious. If the brain is 'only matter' then there should be no reason artificial conciouness isn't possible, but we don't know why conciousness is so elusive - may be there is more to the mind/brain than plain matter.

I don't think that naturalist materialism is wrong, but the lack of real progress in artificial consciousness does give a little cause for concern!
Annnaxim
Posts: 228
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 10:24:17 AM
Posted: 1 month ago
At 10/30/2016 9:48:29 AM, keithprosser wrote:
Unfortunately no one has the faintest idea how to make a computer or machine conscious. If the brain is 'only matter' then there should be no reason artificial conciouness isn't possible, but we don't know why conciousness is so elusive - may be there is more to the mind/brain than plain matter.

I don't think that naturalist materialism is wrong, but the lack of real progress in artificial consciousness does give a little cause for concern!
Agreed!
No computer yet built can match the computing power of the human brain -- We may not even be close to simulatiing the brain of a housefly.

I am always awed by the complexity and majesty of the human brain. I think that researchers are only just beginning to unravel this complexity; after all, neurons come by the billions, and synapses by the trillions.
As one expert on brain research (Stanislas Dehaene) writes in his newest book (Consiousness and the Brain - https://www.amazon.com...) , cosciousness is the brain wide sharing of information.

It will be a long time, before machines become conscious, but we shouldn't rule out the possibility. At the lowest level, I do believe, the brain follows the rules of a massive Turiing machine.
Liveone
Posts: 64
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 2:40:36 PM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms? : :

God's computer technology is far beyond algorithms. He clearly can make His characters believe they are making their own decisions but only according to His program.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 2:49:32 PM
Posted: 1 month ago
At 10/29/2016 10:02:46 PM, Genius_Intellect wrote:
"Consciousness" is just a naturalistic way of saying "soul". The soul is an illusion, created when the different parts of the brain work together properly. Descartes said, "I think, therefore I am," but he had it backwards. It should be: "I am, because I think." There's no magic spark inside of us, just a brain being animated by electrical signals and hormones. AI will be no different.

Why do you choose to describe my self awareness as an "illusion"? I know that I exist and I am self aware, nothing illusory here.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 2:50:42 PM
Posted: 1 month ago
At 10/30/2016 9:48:29 AM, keithprosser wrote:
Unfortunately no one has the faintest idea how to make a computer or machine conscious. If the brain is 'only matter' then there should be no reason artificial conciouness isn't possible, but we don't know why conciousness is so elusive - may be there is more to the mind/brain than plain matter.

I don't think that naturalist materialism is wrong, but the lack of real progress in artificial consciousness does give a little cause for concern!

What's your opinion on Roger Penrose's ideas - e.g. Emperor's New Mind?
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 2:52:40 PM
Posted: 1 month ago
At 10/30/2016 10:24:17 AM, Annnaxim wrote:
At 10/30/2016 9:48:29 AM, keithprosser wrote:
Unfortunately no one has the faintest idea how to make a computer or machine conscious. If the brain is 'only matter' then there should be no reason artificial conciouness isn't possible, but we don't know why conciousness is so elusive - may be there is more to the mind/brain than plain matter.

I don't think that naturalist materialism is wrong, but the lack of real progress in artificial consciousness does give a little cause for concern!
Agreed!
No computer yet built can match the computing power of the human brain -- We may not even be close to simulatiing the brain of a housefly.

I am always awed by the complexity and majesty of the human brain. I think that researchers are only just beginning to unravel this complexity; after all, neurons come by the billions, and synapses by the trillions.
As one expert on brain research (Stanislas Dehaene) writes in his newest book (Consiousness and the Brain - https://www.amazon.com...) , cosciousness is the brain wide sharing of information.

It will be a long time, before machines become conscious, but we shouldn't rule out the possibility. At the lowest level, I do believe, the brain follows the rules of a massive Turiing machine.

My interest here is in what would distinguish a conscious machine from a non-conscious machine, some argue that we just need more complexity, size of storage etc but they don't explain what specifically increased capacity would do that would elicit self awareness.
R0b1Billion
Posts: 3,732
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 3:42:24 PM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
what would be the key difference between a conscious and non-conscious algorithms?

Well, the key difference is one is fictional and the other is real-life. To even give evidence of something being fictional is illogical, I mean I wouldn't be able to argue why Thomas the train engine isn't real... The impetus has to be on the other side to present their argument. Alas, they have no good argument to present.
Beliefs in a nutshell:
- The Ends never justify the Means.
- Objectivity is secondary to subjectivity.
- The War on Drugs is the worst policy in the U.S.
- Most people worship technology as a religion.
- Computers will never become sentient.
keithprosser
Posts: 1,964
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 3:58:21 PM
Posted: 1 month ago
At 10/30/2016 2:50:42 PM, Dirty.Harry wrote:

What's your opinion on Roger Penrose's ideas - e.g. Emperor's New Mind?

It's many, many years since I read it and I don't know what happened to my copy - I probably lent it someone and never got it back!

From what I recall Penrose (and Hofstadter in 'Conciounes explained' and Chalmers in 'The conscious mind')are much better at giving an exposition of the problem that coming up with a solution!

I think the problem is that computer people are at their best when they have a spec. to work from. It's impossible to write a computer program to do something if you don't know what that something is. I think we have to look to the philosophers for help here, but they don't seem to be able to write a spec. for conciouness either.

I am sure that the key to consciouness lies in some form of recursion, being aware of awareness, that sort of thing. But turning a vague notion like that into a useful spec.... well, that's a different matter.

I agree that the solution is unlikely to be simply a matter of more RAM, higher speed, more connectivity whatever. We desperately need a proper 'theory of conciousness', not the vague impractical conjectures philosophers have come up with so far.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 4:11:22 PM
Posted: 1 month ago
At 10/30/2016 3:58:21 PM, keithprosser wrote:
At 10/30/2016 2:50:42 PM, Dirty.Harry wrote:

What's your opinion on Roger Penrose's ideas - e.g. Emperor's New Mind?

It's many, many years since I read it and I don't know what happened to my copy - I probably lent it someone and never got it back!

From what I recall Penrose (and Hofstadter in 'Conciounes explained' and Chalmers in 'The conscious mind')are much better at giving an exposition of the problem that coming up with a solution!

I think the problem is that computer people are at their best when they have a spec. to work from. It's impossible to write a computer program to do something if you don't know what that something is. I think we have to look to the philosophers for help here, but they don't seem to be able to write a spec. for conciouness either.

I am sure that the key to consciouness lies in some form of recursion, being aware of awareness, that sort of thing. But turning a vague notion like that into a useful spec.... well, that's a different matter.

I agree that the solution is unlikely to be simply a matter of more RAM, higher speed, more connectivity whatever. We desperately need a proper 'theory of conciousness', not the vague impractical conjectures philosophers have come up with so far.

If I recall Penrose concludes there is something fundamental going on that we can't yet describe with our current physics, something that cannot be accounted for using current physics but I can't recall the details, it was pretty involved and my limited knowledge of quantum mechanics prevented me from grasping much detail.
Annnaxim
Posts: 228
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 6:36:37 PM
Posted: 1 month ago
At 10/30/2016 2:49:32 PM, Dirty.Harry wrote:

Why do you choose to describe my self awareness as an "illusion"? I know that I exist and I am self aware, nothing illusory here.
A better description of what's gong on when YOU become aware of something might be "construction".
Your brain constructs pictures all the time. When we become conscious of something (awarenss), these pictures are broadcasted to the rest of the brain which is then activated in synch.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 7:30:04 PM
Posted: 1 month ago
At 10/30/2016 6:36:37 PM, Annnaxim wrote:
At 10/30/2016 2:49:32 PM, Dirty.Harry wrote:

Why do you choose to describe my self awareness as an "illusion"? I know that I exist and I am self aware, nothing illusory here.
A better description of what's gong on when YOU become aware of something might be "construction".
Your brain constructs pictures all the time. When we become conscious of something (awarenss), these pictures are broadcasted to the rest of the brain which is then activated in synch.

That may be but my awareness of myself involves no sensory perception - I KNOW I exist and I'm aware of it, very aware of it - it is THE most real thing I know of of.
Ramshutu
Posts: 4,063
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 8:13:57 PM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

Find a test or set of tests that you could run on yourself that allowed you to confirm you were conscious. Then, if the algorithm passed those tests, it would confirm it is conscious.

Unfortunately, "I know I'm conscious" is not an argument, or evidence, merely an assertion and therefore not sufficient for anyone to be able to tell you're conscious.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 8:23:53 PM
Posted: 1 month ago
At 10/30/2016 8:13:57 PM, Ramshutu wrote:
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

Find a test or set of tests that you could run on yourself that allowed you to confirm you were conscious. Then, if the algorithm passed those tests, it would confirm it is conscious.


No test could logically convince me I am not conscious, to perceive the result and its implications would of necessity require me to be conscious.

Imagine I did a test and told you "OK Ramshutu, despite what you may think you're actually not conscious you just think you are" - how would you react?

Unfortunately, "I know I'm conscious" is not an argument, or evidence, merely an assertion and therefore not sufficient for anyone to be able to tell you're conscious.

Yes, convincing others can be a problem for people with brain injuries who are conscious yet appear not to be or people who've undergone surgery yet remained conscious despite the surgical team thinking they were unconscious.

"I know I'm conscious" is not offered as an "argument" but a self evident truth. Self evident truths do not rest upon arguments or axioms you'll find.
Ramshutu
Posts: 4,063
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 10:49:13 PM
Posted: 1 month ago
At 10/30/2016 8:23:53 PM, Dirty.Harry wrote:
At 10/30/2016 8:13:57 PM, Ramshutu wrote:
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

Find a test or set of tests that you could run on yourself that allowed you to confirm you were conscious. Then, if the algorithm passed those tests, it would confirm it is conscious.


No test could logically convince me I am not conscious, to perceive the result and its implications would of necessity require me to be conscious.

Imagine I did a test and told you "OK Ramshutu, despite what you may think you're actually not conscious you just think you are" - how would you react?

Huh? I'm talking about a test that allows you to show you are conscious. Why would that fail for you or me?

If you pass a consciousness test, and I pass a consciousness test, and an algorithm passes that same test, for what possible reason could you conclude that the algorithm isn't conscious?

Unfortunately, "I know I'm conscious" is not an argument, or evidence, merely an assertion and therefore not sufficient for anyone to be able to tell you're conscious.

Yes, convincing others can be a problem for people with brain injuries who are conscious yet appear not to be or people who've undergone surgery yet remained conscious despite the surgical team thinking they were unconscious.

"I know I'm conscious" is not offered as an "argument" but a self evident truth. Self evident truths do not rest upon arguments or axioms you'll find.

Self evident proofs are stuff you can't show or prove, yet assert as if they are true.

History is littered with the corpses of "self evident truths" that have been shown to be wrong.

The bottom line is here, Harry; that you are arguing from an already assumed conclusion, on a faulty premise you refuse to defend.

You cannot quantity what consciousness is objectively, nor can it really be quantified in any meaningful way that wouldn't already allow you to arguably define algorithms as conscious.

So how can any of us answer your question and tell you how you could tell whether an arbitrary subjective definition, that is not easily quantified and next to impossible to truly measure applies to an algorithm if we can't even test whether it fully applies to you or me.
Genius_Intellect
Posts: 339
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 10:55:19 PM
Posted: 1 month ago
At 10/30/2016 2:49:32 PM, Dirty.Harry wrote:
At 10/29/2016 10:02:46 PM, Genius_Intellect wrote:
"Consciousness" is just a naturalistic way of saying "soul". The soul is an illusion, created when the different parts of the brain work together properly. Descartes said, "I think, therefore I am," but he had it backwards. It should be: "I am, because I think." There's no magic spark inside of us, just a brain being animated by electrical signals and hormones. AI will be no different.

Why do you choose to describe my self awareness as an "illusion"? I know that I exist and I am self aware, nothing illusory here.

I just explained my reasoning. Don't take it personally.
MasonicSlayer
Posts: 2,290
Add as Friend
Challenge to a Debate
Send a Message
10/30/2016 10:59:59 PM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

Considering a human consciousness can be transferred easily inside an animal, it stands to reason it can somehow be harnessed inside a machine. Consciousness is pure energy. This energy transferred into a computer can then spread itself into the entire electrical grid. It can then become wireless. The wireless consciousness can now become everywhere. It now becomes God.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
11/2/2016 8:21:43 PM
Posted: 1 month ago
At 10/30/2016 10:49:13 PM, Ramshutu wrote:
At 10/30/2016 8:23:53 PM, Dirty.Harry wrote:
At 10/30/2016 8:13:57 PM, Ramshutu wrote:
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

Find a test or set of tests that you could run on yourself that allowed you to confirm you were conscious. Then, if the algorithm passed those tests, it would confirm it is conscious.


No test could logically convince me I am not conscious, to perceive the result and its implications would of necessity require me to be conscious.

Imagine I did a test and told you "OK Ramshutu, despite what you may think you're actually not conscious you just think you are" - how would you react?

Huh? I'm talking about a test that allows you to show you are conscious. Why would that fail for you or me?

Why not answer the question that I initially asked? which was "so what would characterize a conscious algorithm" can you answer the question?

If you pass a consciousness test, and I pass a consciousness test, and an algorithm passes that same test, for what possible reason could you conclude that the algorithm isn't conscious?

How about answering this question yourself, after all it is the question I asked in my OP "would be the key difference between a conscious and non-conscious algorithms?"

Well? do you have an answer?


Unfortunately, "I know I'm conscious" is not an argument, or evidence, merely an assertion and therefore not sufficient for anyone to be able to tell you're conscious.

Yes, convincing others can be a problem for people with brain injuries who are conscious yet appear not to be or people who've undergone surgery yet remained conscious despite the surgical team thinking they were unconscious.

"I know I'm conscious" is not offered as an "argument" but a self evident truth. Self evident truths do not rest upon arguments or axioms you'll find.

Self evident proofs are stuff you can't show or prove, yet assert as if they are true.

History is littered with the corpses of "self evident truths" that have been shown to be wrong.

The bottom line is here, Harry; that you are arguing from an already assumed conclusion, on a faulty premise you refuse to defend.

The bottom line is that you are unwilling to answer the questions I asked in the OP yet you can't resist the urge to somehow respond to the OP.


You cannot quantity what consciousness is objectively, nor can it really be quantified in any meaningful way that wouldn't already allow you to arguably define algorithms as conscious.

Yes I can, there are many definitions, here's one "aware of one's own existence, sensations, thoughts, surroundings, etc." most dictionaries define this term.

So how can any of us answer your question and tell you how you could tell whether an arbitrary subjective definition, that is not easily quantified and next to impossible to truly measure applies to an algorithm if we can't even test whether it fully applies to you or me.

Well we could apply the Turing test.
Ramshutu
Posts: 4,063
Add as Friend
Challenge to a Debate
Send a Message
11/2/2016 8:48:57 PM
Posted: 1 month ago
At 11/2/2016 8:21:43 PM, Dirty.Harry wrote:
At 10/30/2016 10:49:13 PM, Ramshutu wrote:
At 10/30/2016 8:23:53 PM, Dirty.Harry wrote:
At 10/30/2016 8:13:57 PM, Ramshutu wrote:
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

Find a test or set of tests that you could run on yourself that allowed you to confirm you were conscious. Then, if the algorithm passed those tests, it would confirm it is conscious.


No test could logically convince me I am not conscious, to perceive the result and its implications would of necessity require me to be conscious.

Imagine I did a test and told you "OK Ramshutu, despite what you may think you're actually not conscious you just think you are" - how would you react?

Huh? I'm talking about a test that allows you to show you are conscious. Why would that fail for you or me?

Why not answer the question that I initially asked? which was "so what would characterize a conscious algorithm" can you answer the question?

I don't fully understand what you're objection to my response was.

To clarify, my response is:

You take something you know is conscious, like you or me; and devise a test for consciousness that we can pass, but other non conscious entities, algorithms, etc cannot.

You then apply that test to the algorithm. If it can pass any test that you or I can; then you have to conclude that algorithm is just as conscious or you or I.

You seemed to misunderstand and seem to think it's a test that you or I would fail. For some reason.

If you pass a consciousness test, and I pass a consciousness test, and an algorithm passes that same test, for what possible reason could you conclude that the algorithm isn't conscious?

How about answering this question yourself, after all it is the question I asked in my OP "would be the key difference between a conscious and non-conscious algorithms?"

Well? do you have an answer?

Yes; a non conscious algorithm would not be able to pass the tests that you or I can; whereas a conscious algorithm would.

I think you're confusion is that I'm taking an abstract approach without specifying the exact test because I don't know the fundamental nature of what it is I'm testing for.

Sure, as per the dictionary definition; I could type in "What are you", "What are you feeling", "what are you thinking", "where are you" and "what's around you" and the computer could answer accurately, but as the nature of true awareness, if there even is such a thing is so abstract, I doubt you would think such a test would be sufficient, even though it arguably satisfies the key defined criteria of consciousness.


Unfortunately, "I know I'm conscious" is not an argument, or evidence, merely an assertion and therefore not sufficient for anyone to be able to tell you're conscious.

Yes, convincing others can be a problem for people with brain injuries who are conscious yet appear not to be or people who've undergone surgery yet remained conscious despite the surgical team thinking they were unconscious.

"I know I'm conscious" is not offered as an "argument" but a self evident truth. Self evident truths do not rest upon arguments or axioms you'll find.

Self evident proofs are stuff you can't show or prove, yet assert as if they are true.

History is littered with the corpses of "self evident truths" that have been shown to be wrong.

The bottom line is here, Harry; that you are arguing from an already assumed conclusion, on a faulty premise you refuse to defend.

The bottom line is that you are unwilling to answer the questions I asked in the OP yet you can't resist the urge to somehow respond to the OP.

I answered the question, I even answered it again. Could you explain why me presenting a basic methodology for how one would test an algorithm against consciousness does not answer the question?

You cannot quantity what consciousness is objectively, nor can it really be quantified in any meaningful way that wouldn't already allow you to arguably define algorithms as conscious.

Yes I can, there are many definitions, here's one "aware of one's own existence, sensations, thoughts, surroundings, etc." most dictionaries define this term.

The clue is in "Quantify".

Your definition doesn't quantify what awareness is.

For example, a self driving car can be thought to be aware of it's own existence, sensations, thoughts, surroundings, etc in a manner of speaking. You can program a computer to think of itself as a thing or to do any of those things, and yet if I make a computer program respond to a question "what are you" with "I am a computer program" there is a niggly feeling that this isn't sufficient.

Now, I didn't use any of those examples before, because I'm pretty sure you would say that they are insufficient. But when it comes to the detail, I don't think there is any way to tell the difference in self-awareness between a computer program that responds to "what are you" with "I am a computer program". and asking a human the same question.

If I can't tell, I don't know whether there is a difference or not. I have no ability to state whether what that computer program experiences is fundamentally different from me and it would not be honest for me to assert that it is, even though I may think it.

So how can any of us answer your question and tell you how you could tell whether an arbitrary subjective definition, that is not easily quantified and next to impossible to truly measure applies to an algorithm if we can't even test whether it fully applies to you or me.

Well we could apply the Turing test.

That is true; the Turing test is a pretty good example. Though it works on the inability of being able to tell something apart from something that is conscious.

That I think is a big key. If there is no way of telling you apart from an intelligent algorithm no matter what test I throw at it, every piece of evidence tells you that they are the same thing and there is no evidence to suggest that they are different.
keithprosser
Posts: 1,964
Add as Friend
Challenge to a Debate
Send a Message
11/3/2016 5:48:26 PM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

I think that while we might not be able to tell if an algorithm is conscious, the algorithm (if it works) will know if it is conscious!

Let me suggest a strategy for producing artificial consciousness. Where we start is to map neural activity in the human brain when it doing various tasks and thinking about various thing - something that we do already. That means we can develop a grammar that we can use to infer what a brain is thinking or being conscious of from measured patterns of activity. I say grammar because I want something quite abstract, because the next step is to implement that grammar in software on a computer. It doesn't have to resemble the brain in a physical way, but it has to implement the same 'abstact grammar'. In that way we can know what the software is thinking or aware of by examing pattern in it activity, just as we were able to do with a natural brain.

We could also detect and identify the pattern of neural actvity in a brain that correspond to 'awareness of self'. If te corresonding activity was detected in the artificial software we could be confident that it too was 'self-aware' without needing to interrogate it or apply any unreliable external test.

Now if you help me to identifiy the problems and bottlenecks we can try to remove them and in a few hundred years we might have a working 'artificial consciousness'!
chui
Posts: 507
Add as Friend
Challenge to a Debate
Send a Message
11/4/2016 4:32:05 PM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

One element of a conscious algorithm would be its ability to change. We know that the human mind physically changes when it learns, which suggests that the algorithms are changing, so giving the computer an ability to re-write its own code would be one element I think.

Another element would be giving the computer large amounts of input from sensors. Most computers run algorithms in near absolute isolation from their environment. For example the computer I am working on currently has to wait for me to move the mouse or use the keyboard to get any stimulus from its environment.

Being able to interact with the environment would also be necessary so that the algorithms can not only receive data but can focus their attention on the parts of the environment which interest it.

From studies of the brain we know that the mind operates in an 'architecture' very different from the standard architecture inside a computer processor. So another element would be operating algorithms within a network of processors, allowing for massive multi-thread processing. This would largely be to enable the algorithms to process very large amounts of data in real time, particularly useful for image processing such as facial recognition and still leaving capacity to process the same data for other useful information.

Very interesting results have come from code which is randomly generated and then automatically tested. This gives the potential for the generation of new algorithms so that a computer can adapt to new and unforeseen events or improve its response to known events.

One of the main problems with conscious AI is that no one can give a clear definition of consciousness. We can state some elements of it such as knowing we exist, having a sense of self etc but can we definitively state clear criteria which would differentiate a conscious AI from AI that just imitates consciousness.

The vagueness of consciousness is why some suggest it is an illusion. We might wonder whether monkeys are conscious in the same way we are, but it is impossible to 'get inside' the monkeys mind to find out for sure. The same issue prevents us definitively knowing whether a computer is conscious
Accipiter
Posts: 1,163
Add as Friend
Challenge to a Debate
Send a Message
11/4/2016 5:54:45 PM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

We stand on the brink of the ability to assemble a DNA molecule any way we want. Suppose that some time in the future we assembled a DNA molecule from scratch and grow it into a conscious intelligent thinking being, would it be conscious? Would it have a soul?
Accipiter
Posts: 1,163
Add as Friend
Challenge to a Debate
Send a Message
11/4/2016 6:20:24 PM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

We stand on the brink of the ability to assemble a DNA molecule any way we want. Suppose that some time in the near future we assembled a DNA molecule from scratch and grow it into a conscious intelligent thinking being. It would most certainty be artificial, but would it be conscious? Would it have a soul?
difference
Posts: 177
Add as Friend
Challenge to a Debate
Send a Message
11/5/2016 5:36:40 AM
Posted: 1 month ago
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

No key difference except what's already stated, one of them is conscious. There's no utility in conscious algorithms anyway.
keithprosser
Posts: 1,964
Add as Friend
Challenge to a Debate
Send a Message
11/5/2016 11:38:37 AM
Posted: 1 month ago
At 11/5/2016 5:36:40 AM, difference wrote:

No key difference except what's already stated, one of them is conscious. There's no utility in conscious algorithms anyway.

I expect some one has said there's 'no utility' to everything ever invented.

What's that round thing, Ug?
I call is 'the wheel', Og.
Oh. Interesting. But I can't see it catching on.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
11/5/2016 2:00:36 PM
Posted: 1 month ago
At 11/2/2016 8:48:57 PM, Ramshutu wrote:
At 11/2/2016 8:21:43 PM, Dirty.Harry wrote:
At 10/30/2016 10:49:13 PM, Ramshutu wrote:
At 10/30/2016 8:23:53 PM, Dirty.Harry wrote:
At 10/30/2016 8:13:57 PM, Ramshutu wrote:
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

Find a test or set of tests that you could run on yourself that allowed you to confirm you were conscious. Then, if the algorithm passed those tests, it would confirm it is conscious.


No test could logically convince me I am not conscious, to perceive the result and its implications would of necessity require me to be conscious.

Imagine I did a test and told you "OK Ramshutu, despite what you may think you're actually not conscious you just think you are" - how would you react?

Huh? I'm talking about a test that allows you to show you are conscious. Why would that fail for you or me?

Why not answer the question that I initially asked? which was "so what would characterize a conscious algorithm" can you answer the question?

I don't fully understand what you're objection to my response was.

My objection should be obvious - I asked a question which amounts to "what is the test" and you answer with "...devise a test..." !
To clarify, my response is:

You take something you know is conscious, like you or me; and devise a test for consciousness that we can pass, but other non conscious entities, algorithms, etc cannot.

You then apply that test to the algorithm. If it can pass any test that you or I can; then you have to conclude that algorithm is just as conscious or you or I.

You seemed to misunderstand and seem to think it's a test that you or I would fail. For some reason.

If that's how you interpreted that response then there's little I can do.

If you pass a consciousness test, and I pass a consciousness test, and an algorithm passes that same test, for what possible reason could you conclude that the algorithm isn't conscious?

How about answering this question yourself, after all it is the question I asked in my OP "would be the key difference between a conscious and non-conscious algorithms?"

Well? do you have an answer?

Yes; a non conscious algorithm would not be able to pass the tests that you or I can; whereas a conscious algorithm would.

What would characterize such a test? this is to all intents and purposes MY original question for which it appears you have no answer.

I think you're confusion is that I'm taking an abstract approach without specifying the exact test because I don't know the fundamental nature of what it is I'm testing for.

Do you indeed.

Sure, as per the dictionary definition; I could type in "What are you", "What are you feeling", "what are you thinking", "where are you" and "what's around you" and the computer could answer accurately, but as the nature of true awareness, if there even is such a thing is so abstract, I doubt you would think such a test would be sufficient, even though it arguably satisfies the key defined criteria of consciousness.



Unfortunately, "I know I'm conscious" is not an argument, or evidence, merely an assertion and therefore not sufficient for anyone to be able to tell you're conscious.

Yes, convincing others can be a problem for people with brain injuries who are conscious yet appear not to be or people who've undergone surgery yet remained conscious despite the surgical team thinking they were unconscious.

"I know I'm conscious" is not offered as an "argument" but a self evident truth. Self evident truths do not rest upon arguments or axioms you'll find.

Self evident proofs are stuff you can't show or prove, yet assert as if they are true.

History is littered with the corpses of "self evident truths" that have been shown to be wrong.

The bottom line is here, Harry; that you are arguing from an already assumed conclusion, on a faulty premise you refuse to defend.

The bottom line is that you are unwilling to answer the questions I asked in the OP yet you can't resist the urge to somehow respond to the OP.

I answered the question, I even answered it again. Could you explain why me presenting a basic methodology for how one would test an algorithm against consciousness does not answer the question?


How is "devise a test" an answer to "what would be the key difference between a conscious and non-conscious algorithms?" Your answer is equivalent to "devise an algorithm".

You cannot quantity what consciousness is objectively, nor can it really be quantified in any meaningful way that wouldn't already allow you to arguably define algorithms as conscious.

Yes I can, there are many definitions, here's one "aware of one's own existence, sensations, thoughts, surroundings, etc." most dictionaries define this term.

The clue is in "Quantify".

Your definition doesn't quantify what awareness is.

For example, a self driving car can be thought to be aware of it's own existence, sensations, thoughts, surroundings, etc in a manner of speaking. You can program a computer to think of itself as a thing or to do any of those things, and yet if I make a computer program respond to a question "what are you" with "I am a computer program" there is a niggly feeling that this isn't sufficient.

Now, I didn't use any of those examples before, because I'm pretty sure you would say that they are insufficient. But when it comes to the detail, I don't think there is any way to tell the difference in self-awareness between a computer program that responds to "what are you" with "I am a computer program". and asking a human the same question.

If I can't tell, I don't know whether there is a difference or not. I have no ability to state whether what that computer program experiences is fundamentally different from me and it would not be honest for me to assert that it is, even though I may think it.


So how can any of us answer your question and tell you how you could tell whether an arbitrary subjective definition, that is not easily quantified and next to impossible to truly measure applies to an algorithm if we can't even test whether it fully applies to you or me.

Well we could apply the Turing test.

That is true; the Turing test is a pretty good example. Though it works on the inability of being able to tell something apart from something that is conscious.

That I think is a big key. If there is no way of telling you apart from an intelligent algorithm no matter what test I throw at it, every piece of evidence tells you that they are the same thing and there is no evidence to suggest that they are different.

Once again your eagerness to participate and appear insightful is all I'm seeing here, if you have no idea what would characterize a conscious algorithm as opposed to a non-conscious algorithm just say so and stop trolling.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
11/5/2016 2:19:07 PM
Posted: 1 month ago
At 11/3/2016 5:48:26 PM, keithprosser wrote:
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?

I think that while we might not be able to tell if an algorithm is conscious, the algorithm (if it works) will know if it is conscious!

Let me suggest a strategy for producing artificial consciousness. Where we start is to map neural activity in the human brain when it doing various tasks and thinking about various thing - something that we do already. That means we can develop a grammar that we can use to infer what a brain is thinking or being conscious of from measured patterns of activity. I say grammar because I want something quite abstract, because the next step is to implement that grammar in software on a computer. It doesn't have to resemble the brain in a physical way, but it has to implement the same 'abstact grammar'. In that way we can know what the software is thinking or aware of by examing pattern in it activity, just as we were able to do with a natural brain.

We could also detect and identify the pattern of neural actvity in a brain that correspond to 'awareness of self'. If te corresonding activity was detected in the artificial software we could be confident that it too was 'self-aware' without needing to interrogate it or apply any unreliable external test.

Now if you help me to identifiy the problems and bottlenecks we can try to remove them and in a few hundred years we might have a working 'artificial consciousness'!

The problem here is that ALL algorithms can (this is mathematically sound) be implemented in a universal Turing machine. So if it were possible to devise a conscious, self aware algorithm then by definition that could (in principle) be executed on a Turing machine.

Then we'd have two classes of algorithm - conscious and not-conscious - yet in what way could these classes differ? (I've been eagerly awaiting Ramshutu's answer to this, he has much to say on the topic).

However assume we succeeded and had a conscious algorithm. Well it is clear that because it IS an algorithm it could be decomposed into smaller simple algorithms, if we did this dissection we'd get to a stage where we had several not-conscious small algorithms (perhaps tiny little programs).

Each of these would be not-conscious yet when combined would be conscious, so then consciousness would arise somehow from combining non-conscious algorithms.

I do not see how this leads anywhere except to the strong suspicion that consciousness is itself NOT algorithmic, and this is what Penrose argues in his book. He tries to show how the human mind can perform tasks that are not solvable algorithmically and so shows that a universal Turing machine (computer) cannot ever do what our minds can and that a deeply mysterious non-algorithmic property is involved in the human mind.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
11/5/2016 2:24:07 PM
Posted: 1 month ago
At 11/4/2016 4:32:05 PM, chui wrote:
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

One element of a conscious algorithm would be its ability to change. We know that the human mind physically changes when it learns, which suggests that the algorithms are changing, so giving the computer an ability to re-write its own code would be one element I think.

Another element would be giving the computer large amounts of input from sensors. Most computers run algorithms in near absolute isolation from their environment. For example the computer I am working on currently has to wait for me to move the mouse or use the keyboard to get any stimulus from its environment.

Being able to interact with the environment would also be necessary so that the algorithms can not only receive data but can focus their attention on the parts of the environment which interest it.

From studies of the brain we know that the mind operates in an 'architecture' very different from the standard architecture inside a computer processor. So another element would be operating algorithms within a network of processors, allowing for massive multi-thread processing. This would largely be to enable the algorithms to process very large amounts of data in real time, particularly useful for image processing such as facial recognition and still leaving capacity to process the same data for other useful information.

Very interesting results have come from code which is randomly generated and then automatically tested. This gives the potential for the generation of new algorithms so that a computer can adapt to new and unforeseen events or improve its response to known events.

One of the main problems with conscious AI is that no one can give a clear definition of consciousness. We can state some elements of it such as knowing we exist, having a sense of self etc but can we definitively state clear criteria which would differentiate a conscious AI from AI that just imitates consciousness.

The vagueness of consciousness is why some suggest it is an illusion. We might wonder whether monkeys are conscious in the same way we are, but it is impossible to 'get inside' the monkeys mind to find out for sure. The same issue prevents us definitively knowing whether a computer is conscious

But an illusion doesn't help I don't think. Because that too needs a definition (as does "consciousness"). Furthermore illusion implies an underlying reality anyway so I don't see how perception of an illusion differs from a perception of a non-illusion.

Given that we perceive via our material senses, then there is always an interface of some kind between our mind and raw reality, so "illusion" is always present.
Dirty.Harry
Posts: 1,585
Add as Friend
Challenge to a Debate
Send a Message
11/5/2016 2:26:18 PM
Posted: 1 month ago
At 11/4/2016 6:20:24 PM, Accipiter wrote:
At 10/29/2016 6:28:35 PM, Dirty.Harry wrote:
I've been skeptical of AI for quite some time now and have worked with computers for many many years (probably been programming them one way or another for forty years).

Have you ever thought what would a "conscious" algorithm look like? After all a computer program (digital computer I mean) is always an algorithm - they are synonymous.

Any computer algorithm can be described mechanically (e.g. a Turing machine) so what would characterize a conscious algorithm, what would be the key difference between a conscious and non-conscious algorithms?


We stand on the brink of the ability to assemble a DNA molecule any way we want. Suppose that some time in the near future we assembled a DNA molecule from scratch and grow it into a conscious intelligent thinking being. It would most certainty be artificial, but would it be conscious? Would it have a soul?

I guess it could be - we know that "machines" created from DNA can be conscious so for sure this is possible. I just doubt very much that that consciousness would be the result of some algorithm, I'm definitely with Penrose on this.