Here is how to make a computer conscious, self-aware and free willing
1. Camera A: visual input extern -> feeds into 2.
2. Program A: subconsciousness & memory -> feeds into 3.
3. Display A: visual output inner -> feeds into 4.
4. Camera B: visual input inner -> feeds into 5.
5. Program B: consciousness & free will -> feeds into 6.& 2.
6. Speaker: audio output extern
This "being" is quite limited in the ways it can sense and act upon external world. However, I claim it in principle still has all the sufficient hardware to actualize consciousness greater than that of humans, not in qualia sense of the experience, but considering everything else, including free will.
I suppose it might be questioned what exactly should "Program A" be doing and what part of that should go onto inner screen, but most objections I expect to land around "Program B", that it is not what can be called consciousness and that it can not exercise free will. I am interested to hear those arguments. What concepts will you lean on, just how exactly do you disagree?
2. Program A: subconsciousness & memory -> feeds into 3.
3. Display A: visual output inner -> feeds into 4.
4. Camera B: visual input inner -> feeds into 5.
5. Program B: consciousness & free will -> feeds into 6.& 2.
6. Speaker: audio output extern
This "being" is quite limited in the ways it can sense and act upon external world. However, I claim it in principle still has all the sufficient hardware to actualize consciousness greater than that of humans, not in qualia sense of the experience, but considering everything else, including free will.
I suppose it might be questioned what exactly should "Program A" be doing and what part of that should go onto inner screen, but most objections I expect to land around "Program B", that it is not what can be called consciousness and that it can not exercise free will. I am interested to hear those arguments. What concepts will you lean on, just how exactly do you disagree?
Comments (68)
Do you include mirophones to catch sound ? How does your AI learn language? Without it how can it conceptualise and express complex ideas?
Actually this subject interests me too and I have thought about the practicalities of a computer based consciousness. It seemed to me you'd have to synthesise all the attributes of a human. There is no short-cut. Still, I don't want to sound dismissive.. :smile:
Quoting Zelebg
Quoting Zelebg
For the purposes of the intended discussion do you care about how A and B work?
If you create a program that has 'conciouness' and 'free will', then it can be called conciousness and can excercise free will.
How would you achieve that though? If your program is run on a common computer, it will boil down to a deterministic set of instructions. Its audio output will entirely be determined by its initial code and visual input history. The person who codes and interacts with it can have 100% control over its output.
Could you call that free will?
I would argue program can be made to incorporate 'qualia' properties in a sufficiently robust way that enables those concepts to interact with other concepts in the thought function of the consciousness program.
Perhaps computer's inner representation of qualia would be in terms of pie-charts or whatever, but does it really matter if at the end it can still make all the same conclusions and express them with the same kind of semantics as humans do?
I can imagine all my sensory inputs stop working and all I can do is speak. But I would still be able to say I'd rather have a cup of milk than punch in the face.
Is it not sufficient to have goals? If offered choices that are not relevant to those goals, it can then choose at random, which is the only one absolutely free kind of choice. Right?
Oh, I see now where are coming from. I am talking about it in more abstract terms - what can and can not be done in principle. So in this example computer has already learned or has been programmed, and I don't want to go into those detail unless there is an argument it can not be done in principle.
Only if there is an argument such programs could not be made in principle. And I should probably mention both programs are able to modify and expand themselves & each other.
The question of free will, but more generally it's the question of top-down or downward causality. Let me expand this a bit so we have wider range to pick examples from.
Layers of existence - atom - molecule - cell - organ - organism - consciousness - ecosystem - planet - solar system... and there are two important and mysterious boundaries. First, where molecules become 'alive' as a collective in a cell, and second, where organs become 'conscious' as a collective in an organism. But our question stands before any of the layers, and the question is whether these collective entities from higher levels could be something more than just a sum of their parts - is there a point where what actually happens is no longer determined by the dynamics of the lower level elements, but instead by new emergent properties of the higher level?
It all depends on the definition of 'free will'. So I can't answer your question before we settle the definition and the rest of semantics. However, I can claim it is as free as human intention can be, which means determined by such things as personality and goals, if you would agree with this?
How do know that what you have created is conciousness?
Proper definition will be the judge. Agreed?
And by proper I mean the one most of us agree on. But in any case, all arguments put forward here should basically be about some definition or another because we are talking in general terms, limits and possibilities, rather than anything specific. Although examples and comparisons can go into particular details in any of those emergent levels of existence I mentioned.
Since my job here is to show this computer "being" indeed satisfies all the necessary definitions for my claim to be true, then maybe you would like to present the definition of 'consciousness' so we can start?
However you want to define consciousness, im asking how would you know. The reason why Im asking is because it would be very difficult to do, considering how very little we actually know about consciousness. How do you know you will have replicated it in this computer when you would have no way of accounting for missing aspects/basis (because you do not even know what they are)?
Doesn't any one else feel uncomfortable with consciousness already being in this explanation of consciousness?
How is this not a vicious circularity?
It's just description of the functions. It can still be questioned can ordinary computer host such program, or is there something fundamental about those functions symbols can not capture.
You can't just say Newton's laws are not quite complete description of celestial motion for no reason at all, you have to point at something even vaguely, like there is something wrong with Mercury orbit.
You are basing your opinion on some definition of 'consciousness' where there is something unknown about it. What is it? This computer talks, sings, writes poems, knows all the internet and can answer any of your questions on any language at least at the level of Wikipedia standard. It can tell you what it wants, about its personality, its habits, likes, dislikes, wishes, dreams... we can even watch what it dreams. Damn, if it's not more conscious than me, but how much more conscious can it even be?
To rephrase, those are labels not explanations. You may question whether the label is appropriate or not.
Yet how could you know that the system had such an experience?
And if you cannot know such a thing, why doesn't that render the project infertile?
It's like you read that nubered list as arguments, but that's just a list of hardware components. I'm just saying a PC can be programmed to be conscious, self-aware and free willing. I'm not trying to explain anything until someone points to something that needs explaining.
As you were.
Again, how would you every know that a program incorporated qualia?
We can code it into the program and so we can be certain it has it. Can we not? It also has that inner screen, so I can say there it is qualia right there even we can see, unlike my qualia which you can not.
With title I wanted to suggest this particular arrangement of hardware components is important to achieve all that.
Ah, I see, so you think that you can create consciousness by creating a screen for your homunculus!
The errors are compounding here, I think. But I will re-read what you have written, with this new version in mind.
Is that what you have in mind?
Maybe you mean it is fake if computer's inner representation of qualia is, say some list of pie-charts? I'd say no, because your electro-chemical representation of qualia hardly can be expected to make any more sense. And, actual meaning may not be embedded directly in the lower level representation of the concept definition, but calculated relative in connection to all other stored concepts.
This "relative" meaning then may be the same kind of 'feeling' about the same qualia as that of human, even though extracted from different hardware using different symbolization. But I don't think any of it matters if the machine can draw from those concepts, however internally represented, exactly the same conclusions as we do.
See if I can register my objection by giving a simplified version of your machine; one that is intended to understand an utterance by translating it.
1. Microphone A -> feeds into 2.
2. Translator A: feeds into 3.
3. Speaker A: -> feeds into 4.
4. Microphone B: -> feeds into 5.
5. Translator B: -> feeds into 6.& 2.
6. Speaker: audio output extern
So a sentence in English is translated into some other language and then back into English, but included is the reflexive link from 5 to 2.
Would anyone suppose that this device understood the sentences it translated?
I think the situation with your device is the same.
I think it can be argued without it, but yes, little driver seat for consciousness, all with joystick and tiny little monitor so it can play itself as it likes. Isn't that exactly how it feels? It's interesting parallel in any case, and I do not see where the analogy breaks.
Great question. And again to answer it we must first talk about some definition, in this case what "to understand" means. Would you care to define it?
Computers just execute instructions that are really themselves just high-level calls for packaged low-level logic gate operations on bits, the bits themselves only being meaningful to the human observers who assign meaning to them.
You have to explicitly tell the computer what to do at each step. No hand-waving allowed. Suppose you want the computer to feel pain. You can't just write a program that says the following:
if condition X is true, feel pain
How would you go about writing the actual instructions for feeling pain? What are the step by step instructions which, if followed by a machine incapable of feeling, will cause it to feel, and to feel pain?
Let's have it. How to suffer: Step 1...
I don't see why.
Execution of specific functions goes step by step. But what functions will run, with what parameters and when, can be triggered and varies relative not only to external events, but since the process is recursive, change of parameters and function branching may be triggered by the "thought" process itself.
Is pain necessary for consciousness, self-awareness, or free will?
Definition is necessary so we know what exactly is it you are trying to say. Is there some argument I should address? I don't see it.
I've no trust in definitions, nor do I think them needed here.
Is there a sense in which the machine I describe understands? I don't see it. It's just replacing one string with another.
I think that working out a structure for AI in principle is meaningless. You need to consider practice based tests like the Turing test. The advanced use of language is - as far as we humans know - essential for intelligence, so any 'principle' that does not answer how it is to be achieved in practice is suspect.
I am talking about the machine I described. Your marchine needs not to understand, mine does.
It's not because there are arguments it can not be done in principle, on a PC.
I am not aware there are problems around AI learning to speak, are you referring to something specific?
This might be irrelevant, but my only objection is that this sort of thinking assumes the computational theory of mind.
In that case any argument against that theory should work against this machine as well. What is the best criticism for computational theory of mind? And someone would need to argue that, I can't argue against myself. Or can I?
Quoting Zelebg
I really like where you are trying to go, but I believe your premise is not only malformed but misguided. That is, your (circular) question is really asking for a definition of consciousness, and you are trying to imply the AI solution will not require qualia, just agency.
Before I take a stab at it, I'll ask you to clarify a few things:
1. what does 'free will' mean in your program? I suspect you are talking about a sense of agency, and I think that is where your system blows up. Unlike others, I don't expect self-awareness and agency requires qualia; however, I do believe it requires a holistic state of being which you will never get in any kind of conventional coding or AI systems.
2. How will your system be able to know to question/doubt it is consciousness, know if it is communicating with a sentient being or not, and know what any of that really means? I doubt qualia is needed for this line item, but you have to detail how you would do it (and let me shoot you down!) :grin:
3. how do you program it to have an ego "I" in all of its glory and ugliness, which is not what 'free will' is about?
Actually this looks like a visual recognition program more than a conscious computer. Some of the new mobile phones and computers have facial recognition software that will welcome you with a cheery "Hello" when it recognizes your face. And the steps are nearly the same.
1. camera picks up your face
2. camera software sends image to facial recognition software
3. 4. facial recognition software sees and scans image then looks for id in data base
5. facial recognition software finds id and decides to send authorize to user
6. speaker says "Hello"
My cell phone listens to me all the time, I think that most of us know this. I always talk about cats to my phone and it always shows nice pictures of them on the lock screen. But do not try to tell me that the bloody thing is doing it consciously, because that is not true.
You missed @Zelebg's magical module which your cellphone is missing:
5. Program B: consciousness & free will -> feeds into 6.& 2.
He's not going to tell you how he programmed that, though, until you pay the piper, which I think I've done. Now, we're hopping for more than crickets in return..
No, I did not miss it. I explained it.
Consciousness in living beings allows us to recognize things by comparing them to things we know. Once we have recognized a person we can decide to say hello or not.
no. that is pattern recognition, which current AI robots routinely do.
What is the difference?
Let us know when you've figured out how to implement 2 and 5.
No, explain it to me please.
And then maybe you can explain why my telephone is not a conscious thing using Zelebg's method.
IF for some reason nothing happens, try right mouse button click on quote button.
I think Zelebg will explain it to us when he replies. stay tuned...
So you don't have an answer, but I am wrong? :chin:
I already told you where/why I think you are wrong. "5. Program B: consciousness & free will -> feeds into 6.& 2" does not equal "a visual recognition program" and it is Zelebg job to explain to you why not, or your job to explain to us why you logically conclude they are equal or highly similar.
In what way do you think it is different?
Quoting Sir Philo Sophia
But you are the one that said I was wrong, not him.
Quoting Sir Philo Sophia
I did, if you did not understand the comparison just say so.
Why is that a problem?
Most importantly, or for a start, it does not have the same hardware configuration.
Because nobody knows how to implement 2 and 5, or even if it's possible to do so.
Not known is 'what' to implement, so 'how' is not even a question yet. But what I am suggesting here is both what and how to implement, and relevant part is hardware configuration, not software modules.
Ok. How do you implement 2 and 5?
Does not have the same hardware configuration as WHAT?
Your idea of an conscious computer or a human being that has no hardware?
The example I gave you fits exactly to your specs, so what is the problem with it? Can you explain that to me?
Quoting Zelebg
Make up your mind, is it the software or the hardware that you think is important? Because, as the song goes, you can't have one without the other.
My 3. and 4. are hardware, yours software.
Hardware. The first sentence you quoted does not contradict that.
You think the problem is 2. and 5. and I say the so called 'hard problem of consciousness' is 3. and 4., which needs to be implemented with some kind of display / camera system, rather than by any kind of software algorithm.
You said you offered suggestions on how to implement 2 and 5. I didn't see them. Changing the subject doesn't answer my question. How do you implement 2 and 5?
No, I did not say that. I asked you what is the problem, then suggested you do not know what is the problem, and finally I said what is the real problem and how to solve it. Overall, nothing to do with 2 and 5, but 3 and 4.
Someone using your handle wrote:
Quoting Zelebg
I ask again: How do you implement 2 and 5?
You are misinterpreting and I already explained your error.
Do you know how to implement 2 and 5? Yes or no?
Yes.
Are you keeping it secret?
How do you figure that it HAS to be hardware? Why would software need internal a viewing screen for another camera to watch? Why not just go from 2 to 5 without the BS in the middle.
It would make more sense if the external camera feed "showed" (3) the recognition program the data so that it could "see" (4) it. which is exactly as what a facial recognition program does.
Of course he is, but all will be revealed when he goes to pick up his Nobel prize.