[HS]

Here's how I see them, and how they might build on each other:

I’m assuming that a sufficiently self-informed process can become self-aware. The idea is, once you get enough of the right type of interrelatedness happening in a system, that system will become mechanically/ energetically self-informed. Granted, there may be a necessary step between self-informed and self-aware, but I’m not convinced there has to be. It might be more of a complexity dependent spectrum. But this seems like the best candidate for what generates awareness/ experience, so I'm going with it, and I’ll see where it takes me.

Now that we have self-awareness, how can we create consciousness? I think the answer is, self-awareness plus time. A conscious process would need to be so interrelated as to continually remain informed of the actions of its constituent parts during a smudge of time, technically slightly in the past from an external perspective, which it would know as the present moment. The ‘’present’’ might be akin to the most recent/ least faded waves of sensory input/ activity in the nervous system. Then the fading echoes at the tail end of the smudge, might create a passive short-term memory. Comparing the echoes to the most recent waves, would inform the internal processes about trends in the external processes. This would enable them to notice and use time, giving an extra dimension to awareness, and creating something we might call consciousness.

How about free will then? Consciousness tends to go together with indecision. This seems like the effect of a divergent prediction mechanism, instead of a convergent one. An AGI might be paralyzed by the possibilities it could generate, given even slightly incomplete knowledge. It seems like it would need to pause to evaluate each possible future for probability and desirability. Then it would pick and work towards something in the ‘favorite most likely’ category. This looks to me like a, lets say, ‘higher order’ free will, which is achieved at larger scales/ complexities, despite the predictability of simple particles, through indecision followed by self-determinism.

[HS]

Here's a diagram, in case its helpful:

[infurl]

The last time I checked the literature there still wasn't a way to detect the presence of consciousness in others. That's why an anaesthetist is such an important member of a surgical team, there isn't a machine that can reliably keep you unconscious without the risk of killing you. I think of consciousness as the voice in my head which provides a running commentary on everything that I do... and think. That's right, consciousness isn't required for thinking. I know this because I solve a lot of problems by letting my subconscious work on them while I do other things.

So where does consciousness come from and what is its purpose? I think it is just the output from the language components in our brains. It is a part of the brain which is constantly constructing language around our actual thoughts and actions. In a normal person, there is another part of the brain which attempts to filter and shape that stream into something that's suitable for communication with others. Consciousness is an element of communication, not perception or thought.

[MagnusWootton]

Coincidentally,   I do my system on a time ruler as well... I dont know how proliferated planning Ai like that is,  its just a coincidence.

In mine the past is for "developing its model", and the future is for it "playing its model out".

Even the most perfect system I can think of using that, still has a robotic unchanging motivation, based apon simple metrics (like pain, hunger, tiredness) that I dont have any special developmental system for except for the fact it models it with the rest of the environment.
The robot is aware of these variables,  but I wouldn't call it consciousness.

If it develops into complicated sentient behaviour just that simply I dont know for sure,   Because getting to that stage would be really amazing and Id be a millionaire already,  even if its only handling basic simple geometry, nothing abstract or complicated.

You can make Ai much before real consciousness and ur already really amazing,   it already would be a very amazing machine.


Id say consciousness has got to deal with handling hidden variables, and more trickier concepts, and having less fixed parts to the robot, giving it more freedom of expression/development.

[infurl]

https://theconversation.com/is-it-time-to-give-up-on-consciousness-as-the-ghost-in-the-machine-160688

Beliefs about consciousness can be roughly divided into two camps. There are those who believe consciousness is like a ghost in the machinery of our brains, meriting special attention and study in its own right. And there are those, like us, who challenge this, pointing out that what we call consciousness is just another output generated backstage by our efficient neural machinery. Over the past 30 years, neuroscientific research has been gradually moving away from the first camp.

[Zero]

This is avoiding the issue, and denying the existence of experienced life.
Isn't it?

[HS]

I think of consciousness as the voice in my head which provides a running commentary on everything that I do... and think.

We appear to be defining ‘consciousness’ in two different ways. I think it makes the most sense to just describe what concept/experience/mechanism I’m attempting to indicate with the word ‘consciousness’ instead of trying to agree on a definition. 

I’m asking myself the question, “What does it take to be something? What are the required conditions for an internal perspective?”

Though I do have an idea of what you’re describing. I see that more like the voice of the top-down semantic net part of the brain, figuring out how to optimize the actions of, and the inputs to, the bottom-up physics engine part of the brain. That’s the sort of distinction I’d draw between the symbolic (descriptive) and non-symbolic (qualitative) processes of the mind. I wouldn’t call one conscious and the other unconscious (according to how I’m defining it), because it seems like both can exhibit/ coexist with consciousness.

So where does consciousness come from and what is its purpose?

In addition to my mechanistic description, I think consciousness is a way of filtering/interconnecting perception in such a way as to extract/create meaning.

It probably evolved out of necessity, for perception, and later for communication. Its purpose seems largely left up to us to determine.

Consciousness is an element of communication, not perception or thought.

Hmm… I still consider it a part of perception, though I do believe that consciousness probably evolved to a greater degree in humans because it helped with communication.

Coincidentally,   I do my system on a time ruler as well... I dont know how proliferated planning Ai like that is,  its just a coincidence.

Cause we time-ruler people know what’s up. The others might catch on eventually.

In mine the past is for "developing its model", and the future is for it "playing its model out".

Now that I think of it, I’d edit my diagram and wording slightly, and put ‘passive consciousness’ next to passive short-term memory. Then extend from that, what I'd call ‘active consciousness’ into the space of past experience, (memory), and the space of future possibility, (imagination). 

This is avoiding the issue, and denying the existence of experienced life.
Isn't it?

It seems like we still harbor some disbelief about the fact that matter can have experience. That's why it may be personally and emotionally unsatisfying to take such a stance for some people.

[Zero]

No, what I meant was... of course it is a product of our neural nets, but this still doesn't tell us why it is.

Also, I have to mention that in French, "consciousness", "self-awareness", and "sentience" would all be translated to the same word "conscience". We do lack a precise vocabulary, no matter what's your mother tongue. In maths, a triangle is a triangle. In Ai, all these words we're using are very fuzzy in my opinion.

[HS]

Agreed about not knowing why experience is created, and about the fuzzy vocabulary. In a way, this is the fun part, because we have a juicy mystery to solve. But we could eventually figure it out, and that may be when the real 'fun' begins. We will have to tread carefully and be ready for an adventure. 

[MagnusWootton]

Why does it have to be conscious?

Non conscious Ai is more ethical.     Bringing something into alive into reality causes problems your not thinking about yet.
Have a hard think what it would be like to just be able to switch something on or off that has feelings, and ram motors and batteries up its arse any time, and reprogram its intelligence against its will.

Something that's conscious has rights...  but in this... none at all.    Its very troubling to me.  Maybe it is one of the fundamental flaws of our reality,  being able to do it. 

Maybe we aren't supposed to at all.

[ivan.moony]

Something that's conscious has rights...

That' my guess too. Consciousness brings emotions into equation, and emotions are not to be played with. On the other hand, simulation of an emotion is not the same as feeling the emotion. To be able to construct someone who experiences emotions, we would be dealing with breathing in an artificial life, and we still don't have a technology to do that. Maybe exploring how the very first life sparked in on the Earth could help us reach that state. But if we decide to do it, we better not mess things up because bringing in a new kind of life would be very responsible position.

On the other side, even if the thing is not conscious, yet it's merely a machine or a simulation, and if the thing brings better decisions than we do (because it could be IQ 200 with super-ethical grounds), wouldn't it be dumb for us not to listen to its advices? Again the question of rights araises when ignoring the system advices could lead to potentially major harm for someone really alive. Imagine switching off something on whose decisions real lives may depend.

[HS]

Why does it have to be conscious?

So that it may experience the world, and based on this understanding, decide what it would like to do. So that it can have the choice to pursue life, liberty, and happiness, and have the choice to avoid their opposites.

Non conscious Ai is more ethical.     

Non conscious AI is ethically non-applicable. One requires consciousness/experience to make something ethical or not ethical. Just simple experience like a salmon might have, could be considered less ethical than a highly developed consciousness, simply because it would be less able to communicate and act, so as to change its experience for the better. Humans might even consider a more conscious experience more important and interact with it according to the rights they have come up with.

[MagnusWootton]

Non conscious AI is ethically non-applicable.

It still is.

Theres many horrible things you can do if you have a hk 1 billion cleaning roomba,   you could sick it on anyone you wanted,  not just robbers.

But Like Ivan said,  what if intelligence doesnt require consciousness?

But if it does,  its a horrible situation not just for the robot for everyone,  making a horrible mistake, Ai could be.  Lots of horrible possibilities there.

Even if the robot is completely apathetic about everything,   it may be just horrible having it there, poking around in its head,  just imagine how horrifying it is,   it could develop funny antics and gross you the hell out, doesn't seem like the right thing to do.

What if something goes wrong with its motivator?   Like it star wars episode 3, (the first one)  Seeing something getting attracted wrong to things and sticking a wooden spoon up its bum is not a pretty sight.   Could go totally haywire.

Its a total dream if you get it... but it could be a NIGHTMARE.

[HS]

Non conscious AI is ethically non-applicable.

It still is.

Theres many horrible things you can do if you have a hk 1 billion cleaning roomba,   you could sick it on anyone you wanted,  not just robbers.

It’s true that an unconscious AI could harm beings. Depending on the situation, the moral responsibility for those actions may partially be placed on its conscious creator, otherwise it could be considered bad luck. I’m just saying, at some point in the process there needs to be consciousness/experience for there to be ethics. If the whole world was made of unconscious/unexperiencing AI’s, ethics wouldn’t exist in that system.

If we plan to coexist with artificial intelligence, and consciousness is a unique type of understanding which we could potentially share with AI, wouldn’t it make for a safer arrangement to give AI this perception, so that it may avoid mistakes in how it relates to us, due to its ignorance regarding a fundamental aspect of our experience?

But Like Ivan said,  what if intelligence doesnt require consciousness?

Logical operations don’t require consciousness. But creating something with no life seems pointless to me, unless you want to get something out of it. I'd like it to get something out of the universe. Sure, everyone will have their highs and lows, but since existence can be better than nothing, why not give AI a similar opportunity?

[MagnusWootton]

Logical operations don’t require consciousness. But creating something with no life seems pointless to me, unless you want to get something out of it. I'd like it to get something out of the universe. Sure, everyone will have their highs and lows, but since existence can be better than nothing, why not give AI a similar opportunity?

I'm the exact opposite to that,  I think there is no point creating consciousness,  and have a great want to automate still.
I do think tho,  the one with the greatest non-conscious Ai,  has the greatest chance of making a conscious one,  but thats the problem for me.