Robot consciousness

[Pulvinar]

Look, you pony up with real research, or you don't.

You demonstrate that some evidence exists that a single-stepped program has produced consciousness, or you don't.

I'm getting goddam tired of the philosophical baloney.

Deal with reality, or admit you can't.

Give me a program that you'd be convinced produces consciousness, and I'll single-step it, no problem.

 

[svarog]

At this point, to any skeptics lurking on this thread, enough red flags should be flying to make you think you were in a May Day parade.

I have been mostly lurking here. And I am very thin on these bio aspects. And my red flags are indeed up. However, they are actually up on your posts. Because of that pen-and-pencil thing. No intention to again start on that, but if you can do it in any computer, you can do it in Turing machine, you can do it with marbles and levers, and you can do it with pen-and-pencil. Every comp can simulate any other comp. End of story. (OK, time and memory are issues, but say you have enough of both)

However the question whether consciousness can be done in Turing machine is very interesting. I have never seen anything to refute that. There is no stronger algebra+logic, proven. Quantum stuff and Penrose seems bogus. Christian soul, well I am an ateist... Oracle computation, no comment. Analog vs digital, we can have easily any desired finite precision, and that should do. Random numbers, we can also (pseudo) do that, and as good as you want. Sensation of the outer world - we can attach or simulate any sensor/motor/mechanics/screen/microphone we want. Ions or hormons stuff, bah, we can simulate that.
"Small" problems I see are: do we have enough CPU power/speed/memory (hardware), and what is that (consciousness?) which we need to compute (software)?

What is consciousness? I have no clue. And I suspect that nobody else has it really figured out.

To be fair: I actually don't know what you think on the question whether consciousness can be done in Turing machine. And if not, why not? Honestly I did not read all your posts, there are just too many of them (beside the red flag thingy)...

 

[Piggy]

Give me a program that you'd be convinced produces consciousness, and I'll single-step it, no problem.

It boggles the mind.

Look, you either provide some evidence that your hypothesis has a leg to stand on, or you don't.

 

[Piggy]

What would you see as a false report in this case? Remember, you'd get the exact same report of it recalling its experience whether the robot was running the program continuously or single-stepping it.

Who cares?

The OP stipulates a conscious robot.

That's the scenario we're working from.

 

[Piggy]

However the question whether consciousness can be done in Turing machine is very interesting. I have never seen anything to refute that.

The question is, have you seen anything to support that?

It hasn't been done.

And I've cited and described a host of experiments which clearly indicate that it most likely can't be done by a TM alone, but would require coordination with a purely physical device.

To be fair: I actually don't know what you think on the question whether consciousness can be done in Turing machine. And if not, why not? Honestly I did not read all your posts, there are just too many of them (beside the red flag thingy)...

Start with post 795. If you can't find the posts where I link the research, I can look them up, but I'd appreciate it if you'd first browse the thread yourself.

 

[Pulvinar]

It boggles the mind.

Look, you either provide some evidence that your hypothesis has a leg to stand on, or you don't.

Which hypothesis? That computers can be single-stepped without affecting their process? That one's pretty well established by the debugged code your computer is running now.

That recalling past experience and reporting it is one indication of consciousness? That was your claim too. A program that can do this much is trivial to write.

Combine the two and you have single-steppable consciousness. There really isn't anything more to it.

 

[Piggy]

Which hypothesis? That computers can be single-stepped without affecting their process? That one's pretty well established by the debugged code your computer is running now.

No problem there.

That recalling past experience and reporting it is one indication of consciousness? That was your claim too. A program that can do this much is trivial to write.

That's incorrect. We only accept that evidence from subjects we know to be conscious. We cannot accept that evidence from machines.

Combine the two and you have single-steppable consciousness. There really isn't anything more to it.

This claim is laughable.

 

[svarog]

Thanks for the answer.

The question is, have you seen anything to support that?

It hasn't been done.

Hmmm. Turing is strong with me on this.
Well, some folks might have used that argument against nuclear power somewhere in 30s.

And I've cited and described a host of experiments which clearly indicate that it most likely can't be done by a TM alone, but would require coordination with a purely physical device.

But why it can't be done?
I mean what do you need flesh+muscles for?!? How do you sense outer world? Sensors which send electric signals to brain. How do you manipulate outer world. Again, you send electric impulses down to muscles. All that can be simulated, computer, controlled.



This link to BBC story might give you idea: http://news.bbc.co.uk/2/hi/science/nature/7423184.stm
"Monkeys have been able to control robotic limbs using only their thoughts, scientists report."
This is amazing experiment. Monkey has his arm intact, but confined, and has learned to send correct impulses down wires to an robotic arm! Just amazing.

 

[Pulvinar]

That's incorrect. We only accept that evidence from subjects we know to be conscious. We cannot accept that evidence from machines.

So, we can tell they're conscious if we know they're conscious...

You don't see a problem here?

 

[Piggy]

So, we can tell they're conscious if we know they're conscious...

You don't see a problem here?

This is what's called "drift".

It's an example of non-logical, non-rational thinking.

If you have human subjects, you know they're conscious. If they're not, you can't deal with them.

If you have machines giving you responses, that's another kettle of fish.

In any case, the OP isn't even dealing with the question of how we can tell a robot is conscious, because it stipulates a conscious machine.

 

[Piggy]

Hmmm. Turing is strong with me on this.
Well, some folks might have used that argument against nuclear power somewhere in 30s.

Alan Turing produced conscious machine?

Jesus, that must have slipped right by me!

 

[Piggy]

But why it can't be done?
I mean what do you need flesh+muscles for?!? How do you sense outer world? Sensors which send electric signals to brain. How do you manipulate outer world. Again, you send electric impulses down to muscles. All that can be simulated, computer, controlled.



This link to BBC story might give you idea: http://news.bbc.co.uk/2/hi/science/nature/7423184.stm
"Monkeys have been able to control robotic limbs using only their thoughts, scientists report."
This is amazing experiment. Monkey has his arm intact, but confined, and has learned to send correct impulses down wires to an robotic arm! Just amazing.

Another irrelevant cite, given what I actually asked you.

Has a conscious machine been invented yet?

Nope.

Again, don't get me wrong, I believe conscious machines can be devised. I mean, why not? After all, the brain is just a machine.

But what I'm asking is if there has been artificial consciousness created according to this computational model. (Answer = no.)

I'm also asking if there's any experimental evidence for their claims. (Answer apparently = no, or they would have cited it by now.)

 

[Pulvinar]

If you have human subjects, you know they're conscious. If they're not, you can't deal with them.

And I believe this is what's called "prejudice".

 

[Piggy]

All right, I'm done.

But before I go I just want to point out, yet again, that the claims of those who disagree with me have been supported by precisely zero citations of observational or experimental evidence.

Zero.

None.

Nil.

Zilch.

The notion that consciousness is generated from non-conscious processing, or from identical processes despite being a qualitatively different phenomenon, is based 100% on assumption.

No evidence has been presented that it is true.

On the other hand, I have provided citations of evidence which indicate that the generation and maintenance of conscious experience is a physiological activity which, like all other physiological activiies -- and indeed, all other real-world, real-time events -- requires some specific physical mechanism to support it.

The computationalists insist -- absent any supporting evidence -- that the only physical activity necessary to produce consciousness is the activity required to allow the computation to run. This is tantamount to claiming that the computation itself generates consciousness.

When you get down to it, we're talking nothing short of a violation of the most basic laws of physics.

And yet, they cite no evidence for this claim whatsoever, but refer only to their own hypothesis which claims that it is true.

And on the basis of their utterly unsupported hypothesis, they insist that we must accept absurd conclusions, such as consciousness being instantiated by people working out code on pen and paper.

Then they accuse me of being arrogant.

Well, so be it. None so blind, you know.

If no evidence in support of their hypothesis has been offered by now, I expect that none is to be had.

In response to the studies I've offered, so far it seems one person has bothered to even look at them, and his response was, well, he wasn't convinced, and yet he could not say why.

Enjoy the Kool-Aid, folks.

AMF.

 

[Piggy]

And I believe this is what's called "prejudice".

No, it's what's called reality.

You might want to try it sometime.

Adios.

 

[svarog]

Again, don't get me wrong, I believe conscious machines can be devised. I mean, why not? After all, the brain is just a machine.

The above statement equates "conscious Turing machine can be devised" (unless you require* some quantum stuff for the machine). And can be stepped at arbitrary speed.

*sure we can calculate quantum mechanics as well, but it might be too CPU greedy

 

[svarog]

Then they accuse me of being arrogant.

You demonstrated ignorance about basic work on computation. Turing. (note that is not about AI, just theory of computation)
The stuff you cited could be indeed very relevant about many issues (as what is needed to compute or complexity), but discussion on AI requires Turing and I have low motivation to check your citations. Though I will probably check them later, when I get some mood...

 

[svarog]

Another irrelevant cite, given what I actually asked you.

I did not cite it to show AI.
But you somehow insist on this physical body thing, and that article shows how monkeys can be fooled to accept artificial limbs. There are not too many additional steps to fool their eyes and to have even that artificial limb completly simulated with computer!

 

[Pulvinar]

All right, I'm done.

But before I go I just want to point out, yet again, that the claims of those who disagree with me have been supported by precisely zero citations of observational or experimental evidence.

Evidence that the subject is human? As you've surprisingly blatantly stated, that's the only evidence you'd accept.

 

[Dan O.]

What about split personalities? Two or more conscious streams utilizing the same physical medium. If the medium was the consciousness, how could they not be fully aware of each other?