Could a Robot be Conscious?

Discussion of articles that appear in the magazine.

Moderators: AMod, iMod

Science Fan
Posts: 843
Joined: Fri May 26, 2017 5:01 pm

Re: Could a Robot be Conscious?

Post by Science Fan »

I don't know how anyone can conclude that a computer or robot can become conscious? After all, how does one know that there is not something about neurons in an actual biological brain that can give rise to consciousness whereas purely mechanical electronic circuits cannot do so? I'm not saying that is the case, just pointing out that this is presently an unknown and I don't know how one can rule it out without looking at the empirical evidence.
commonsense
Posts: 5116
Joined: Sun Mar 26, 2017 6:38 pm

Re: Could a Robot be Conscious?

Post by commonsense »

Noax wrote: Mon Apr 02, 2018 3:48 pm You seem to have a very anthropomorphic view of robots, possibly due to the Asimov background. Actual robots have a lot of variance, many incapable of locomotion. All seem to be programmed with the 2nd law (perform your function) but otherwise have little to no programming for the other two. Self-driving cars are an exception since they employ all three laws, still giving priority to the second. They have, as do most robots, very weak AI, being programmed and not taught.

'Muscles', 'organs' and 'urinate' seem to be strictly biological terms, and only biological things might have them. Robots might have parallel equivalents, but it would seem wrong to use these words just like it is wrong to use 'gills' to describe my lungs. 'Senses' is a different story. Yes, almost all robots (even a mouse trap) has senses, which is any device capable of gather input. I would hesitate to use the words 'taste' and 'smell' to describe non-biological chemical testing. Grey area.

Yes, they could [display emotions] if their purpose was to imitate a human. Few robots would have that purpose, and none to date are indistinguishable. The Turing test is about being indistinguishable in this way, not about surpassing us in AI capability.

I have a very loose definition of consciousness. It's all about one's definition. I think some very trivial things are conscious, requiring only that it have sensory ability (without which there would be nothing to be conscious of), and ability to react in some appropriate manner to that sensory input. That's really loose.

Many people simply define it as "sufficiently like me", and a robot will never be that, however proficient it might be at imitating us. But then as I pointed out before, more highly evolved aliens would not meet that definition, so the purpose of it seems to be anthropocentric.
Thank you for your post. You've responded to my post in such a way as to add clarity. I disagree with nothing you said. Not that it matters, I would just like to say that I was thinking of an abstract robot that might exist some day.
jayjacobus
Posts: 1273
Joined: Wed Jan 27, 2016 9:45 pm

Re: Could a Robot be Conscious?

Post by jayjacobus »

A robot doesn't experience anything. It compiles lightwaves, sound waves, odor molecules, etc. and then looks up names for the objects and then looks up the definitions. If the robot determines what a lemon smells like and tastes like it is because the definitions tell it but it can't taste or smell. Neither can it hear or see. It fools the naive person by using definitions to make it seem like it does.
commonsense
Posts: 5116
Joined: Sun Mar 26, 2017 6:38 pm

Re: Could a Robot be Conscious?

Post by commonsense »

jayjacobus wrote: Mon Apr 02, 2018 6:42 pm A robot doesn't experience anything. It compiles lightwaves, sound waves, odor molecules, etc. and then looks up names for the objects and then looks up the definitions. If the robot determines what a lemon smells like and tastes like it is because the definitions tell it but it can't taste or smell. Neither can it hear or see. It fools the naive person by using definitions to make it seem like it does.
I was fascinated by your post. I was intrigued to wonder how these robot processes differ from what happens in human beings. Mind you, I am not attempting to refute your statements. I am pondering them, considering their implications and describing my reactions. Perhaps you will straighten things out for me.

In humans, light waves excite receptors on the retina, which causes electrical impulses to travel along the optic nerve to the sensory interpretation areas in the cerebrum in the brain.

Sound waves vibrate structures in the middle ear which, in turn, excite receptors in the inner ear, causing electrical signals to travel along the auditory nerve to the interpretation areas in the cerebrum.

Taste and smell excite chemical receptors on the tongue and in the nasal epithelium, sending impulses along the 1st and 9th cranial nerves to the sensory interpretation areas.

In like fashion, modalities of touch begin with the excitation of neuro-receptors and culminate in the cerebrum.

It is the job of the sensory interpretation areas in the cerebrum to make sense of the electrical signals it receives. Interpretation is done by looking up definitions and making associations from the impulses that are received.

I think it is reasonable to say that robots do not experience. I wonder by what definition are robots excluded from having experience while humans experience the world around them. Perhaps you can help me resolve this.
commonsense
Posts: 5116
Joined: Sun Mar 26, 2017 6:38 pm

Re: Could a Robot be Conscious?

Post by commonsense »

Science Fan wrote: Mon Apr 02, 2018 4:16 pm I don't know how anyone can conclude that a computer or robot can become conscious? After all, how does one know that there is not something about neurons in an actual biological brain that can give rise to consciousness whereas purely mechanical electronic circuits cannot do so? I'm not saying that is the case, just pointing out that this is presently an unknown and I don't know how one can rule it out without looking at the empirical evidence.
I wonder as well about the differences between electrical impulses propagated by neurons in the human nervous system and electrical impulses propagated by circuits in a mechanical system.
jayjacobus
Posts: 1273
Joined: Wed Jan 27, 2016 9:45 pm

Re: Could a Robot be Conscious?

Post by jayjacobus »

Let me explain one sense.
commonsense wrote: Mon Apr 02, 2018 11:04 pm
In humans, light waves excite receptors on the retina, which causes electrical impulses to travel along the optic nerve to the sensory interpretation areas in the cerebrum in the brain.
In humans, light waves excite receptors on the retina, which causes electrical impulses to travel along the optic nerve to the sensory formation in the cerebrum in the brain. The sensory information is perceived and interpreted by consciousness.

It is the job of Consciousness to make sense of the senses it receives. Interpretation is done with or without definitions because you know what you see without labels or definitions.

commonsense wrote: Mon Apr 02, 2018 11:04 pm
I think it is reasonable to say that robots do not experience. I wonder by what definition are robots excluded from having experience while humans experience the world around them. Perhaps you can help me resolve this.
Perception, awareness, interpretation, knowing, understanding, etc.
Londoner
Posts: 783
Joined: Sun Sep 11, 2016 8:47 am

Re: Could a Robot be Conscious?

Post by Londoner »

commonsense wrote: Mon Apr 02, 2018 11:04 pm It is the job of the sensory interpretation areas in the cerebrum to make sense of the electrical signals it receives. Interpretation is done by looking up definitions and making associations from the impulses that are received.
This suggests the 'homunculus'. That our body is no different from a robot's; responding to outside events in the same mechanical way that my keyboard responds to my pressing a key. But sitting inside this machine is a little man who is giving meaning to these purely mechanical events, those electrical signals, just as you are now interpreting the marks on your computer screen.

But how could you do this? Every electrical signal is just that, it doesn't mean anything. The homunculus must have been created with a set of code books. The following electrical signals mean 'a chair'. But that does not help since the looking-up process is still mechanical. We need another homunculus that interprets the result, what 'a chair' means. And so on, in an infinite regression.

I would argue that this doesn't describe what it is for humans to experience or be conscious. We do not get a series of sensory inputs which we process, to then get a particular output, as the computer processes my strokes on the keyboard. Rather our position is that we are alive, that is we are already situated within a world. We do not distinguish inputs from interpretations, not is any interpretation final or discrete from everything else. That any particular interpretation of the world is not exclusive gives us awareness of ourselves, as an arbitrator between possibilities.

So I would say that a robot cannot be given consciousness, certainly not by adding on layer after layer of instructions, homunculus after homunculus. Rather (like humans) it would need to be born into the world, obliged to create its own sensory apparatus and programming as it goes. That seems so contrary to the character of machines that I think the idea that robots could be made to be conscious seems a non-starter.
commonsense
Posts: 5116
Joined: Sun Mar 26, 2017 6:38 pm

Re: Could a Robot be Conscious?

Post by commonsense »

Londoner wrote: Tue Apr 03, 2018 11:00 am
commonsense wrote: Mon Apr 02, 2018 11:04 pm It is the job of the sensory interpretation areas in the cerebrum to make sense of the electrical signals it receives. Interpretation is done by looking up definitions and making associations from the impulses that are received.
This suggests the 'homunculus'. That our body is no different from a robot's; responding to outside events in the same mechanical way that my keyboard responds to my pressing a key. But sitting inside this machine is a little man who is giving meaning to these purely mechanical events, those electrical signals, just as you are now interpreting the marks on your computer screen.

But how could you do this? Every electrical signal is just that, it doesn't mean anything. The homunculus must have been created with a set of code books. The following electrical signals mean 'a chair'. But that does not help since the looking-up process is still mechanical. We need another homunculus that interprets the result, what 'a chair' means. And so on, in an infinite regression.

I would argue that this doesn't describe what it is for humans to experience or be conscious. We do not get a series of sensory inputs which we process, to then get a particular output, as the computer processes my strokes on the keyboard. Rather our position is that we are alive, that is we are already situated within a world. We do not distinguish inputs from interpretations, not is any interpretation final or discrete from everything else. That any particular interpretation of the world is not exclusive gives us awareness of ourselves, as an arbitrator between possibilities.

So I would say that a robot cannot be given consciousness, certainly not by adding on layer after layer of instructions, homunculus after homunculus. Rather (like humans) it would need to be born into the world, obliged to create its own sensory apparatus and programming as it goes. That seems so contrary to the character of machines that I think the idea that robots could be made to be conscious seems a non-starter.
Not to belabor a point, I'm only soliciting your thoughts:

Humans learn to recognize 'chair' to mean chair after seeing lots of chairs and then making a definition of 'chair' by attaching 'chair' to chair. How is this different from the kind of AI where learning takes place by allowing a computer or robot to teach itself definitions as a result of its own inputs?
commonsense
Posts: 5116
Joined: Sun Mar 26, 2017 6:38 pm

Re: Could a Robot be Conscious?

Post by commonsense »

jayjacobus wrote: Tue Apr 03, 2018 7:26 am Let me explain one sense.

In humans, light waves excite receptors on the retina, which causes electrical impulses to travel along the optic nerve to the sensory formation in the cerebrum in the brain. The sensory information is perceived and interpreted by consciousness.

It is the job of Consciousness to make sense of the senses it receives. Interpretation is done with or without definitions because you know what you see without labels or definitions.
This begs the question, what is consciousness, doesn't it?
commonsense wrote: Mon Apr 02, 2018 11:04 pm I wonder by what definition are robots excluded from having experience while humans experience the world around them. Perhaps you can help me resolve this.
jayjacobus wrote: Tue Apr 03, 2018 7:26 am Perception, awareness, interpretation, knowing, understanding, etc.
I think you have well-identified what consciousness does as a result of having experience, but here I meant to ask how experience is defined (yes, I do see my ambiguity about what I wanted to ask). Do you have an idea what distinguishes experience from input?
Londoner
Posts: 783
Joined: Sun Sep 11, 2016 8:47 am

Re: Could a Robot be Conscious?

Post by Londoner »

commonsense wrote: Tue Apr 03, 2018 2:39 pm Humans learn to recognize 'chair' to mean chair after seeing lots of chairs and then making a definition of 'chair' by attaching 'chair' to chair. How is this different from the kind of AI where learning takes place by allowing a computer or robot to teach itself definitions as a result of its own inputs?
I think the difference is that you would have difficulty forming that definition of 'chair'. Our idea of a chair is fuzzy; for example it includes both the idea of 'something you could sit on' but also purpose ' an object made to be sat on'. So a convenient piece of wood might serve as a chair, but not be a chair in the 'purpose' sense. But suppose you bring the piece of wood into your house and use it as a chair? Or you take a purpose made chair into the garden and chop it up for firewood? As Wittgenstein would put it, all these (and more) uses of the word 'chair' are like family resemblances; A resembles B in some way, and B resembles C in some way, but it is hard to find a way C resembles A.

For a human, this does not matter. The word 'chair' is only a tool, it is not so much meaningful as useful. I can ask for a chair and it is understood in the context of the situation; I do not need a precise definition. But the computer is seeking a definition; it needs a specific answer. Unlike me, it does not want to sit down (or do anything) so it doesn't understand words as tools, or actions.
commonsense
Posts: 5116
Joined: Sun Mar 26, 2017 6:38 pm

Re: Could a Robot be Conscious?

Post by commonsense »

OK, OK, I hear you, 1x0, Atla, Brian, Dontaskme, jayjacobus, Londoner, Noax, Science Fan and Walker. Your arguments are powerful and convincing. You have long ago converted me to your reasoning. I continued to address you, though, not to debate you, but to solicit your ideas about questions that troubled me.

I am still troubled by one question that remains.

How come if we are going to assume consciousness in each other, even though we take this assumption without empirical evidence, how come we deny consciousness might exist in robots? After all, robots merely imitate our behavior, yet without empirical evidence we cannot know whether they may or may not have awareness.

I’m just saying, how do we resolve this?
Impenitent
Posts: 4331
Joined: Wed Feb 10, 2010 2:04 pm

Re: Could a Robot be Conscious?

Post by Impenitent »

sitting on a bag filled with beans...

-Imp
User avatar
Noax
Posts: 672
Joined: Wed Aug 10, 2016 3:25 am

Re: Could a Robot be Conscious?

Post by Noax »

commonsense wrote: Tue Apr 03, 2018 8:18 pm How come if we are going to assume consciousness in each other, even though we take this assumption without empirical evidence, how come we deny consciousness might exist in robots?
I for one never denied it.
After all, robots merely imitate our behavior, yet without empirical evidence we cannot know whether they may or may not have awareness.

I’m just saying, how do we resolve this?
Imitation of us is an incredibly poor standard, even if they can do it. Is a cat not conscious just because it isn't capable of fooling anybody into thinking it is human? If a robot is conscious, it will more likely imitate itself. Let it be what it is.

The definition of conscious needs clarification if the issue has a hope of resolution. A dualist might assert that it requires an immaterial mind, in which case he is challenged to demonstrate that even humans have one. A religious person might say cats feel no pain because they don't go to heaven, and thus have no feeling souls.

My definition just says it senses its environment in any way and reacts in some purposeful way to that. Even plants are conscious then. I suppose it needs to be first-person, but it is hard to imagine it not being that. What if it's sensors are wall-mounted? What if the processor isn't where the moving parts are? Is it still first-person? What if it isn't centralized at all? Is a bee hive conscious? If so, is it first person?
Jayjacobus seems to define it as strictly a biological term like 'gill' and thus would be inappropriately used to describe a non-biological mechanism that serves the same purpose. That's fine, but then a different blanket-term needs to be used (something that isn't strictly biological) when asking if robots could have it.
jayjacobus
Posts: 1273
Joined: Wed Jan 27, 2016 9:45 pm

Re: Could a Robot be Conscious?

Post by jayjacobus »

Both the brain and the robot process information. The end results for the brain are senses. The end result for the robot are definitions. Humans get to definitions not through processing but by perceiving the meaning of something. In other words it is not the brain that perceives but consciousness.

Besides I know the meaning of everything I see without naming what I see. I apprehend. The robot processes. Moreover all my sense are always available although my attention is focused. I perceive a whole from which I choose to notice specific items. The robot works from the ground up. For a robot to get to the whole it identifies each separate item.

Biology is a factor because machines aren't conscious.
Londoner
Posts: 783
Joined: Sun Sep 11, 2016 8:47 am

Re: Could a Robot be Conscious?

Post by Londoner »

commonsense wrote: Tue Apr 03, 2018 8:18 pm OK, OK, I hear you, 1x0, Atla, Brian, Dontaskme, jayjacobus, Londoner, Noax, Science Fan and Walker. Your arguments are powerful and convincing. You have long ago converted me to your reasoning. I continued to address you, though, not to debate you, but to solicit your ideas about questions that troubled me.

I am still troubled by one question that remains.

How come if we are going to assume consciousness in each other, even though we take this assumption without empirical evidence, how come we deny consciousness might exist in robots? After all, robots merely imitate our behavior, yet without empirical evidence we cannot know whether they may or may not have awareness.

I’m just saying, how do we resolve this?
I agree, we cannot know that other people have consciousness. See 'P Zombies'. We only posit they do because it explains the various way they behave. But robots do not behave like us, so we have a different explanation as to why robots do what they do. We can say that robots are 'aware' but what would we mean, since we already have a complete explanation for robot behaviour that doesn't need the addition of 'awareness'.

Otherwise, are we discussing a hypothetical robot that behaves just like a human? Then I agree, a robot that behaves exactly like a human would be indistinguishable from a human. But they don't.
Post Reply