Can AI eventually become 'self-aware'?

I say no.

First, let me establish that 'self-aware' is the wrong term. A dog, for example, is not self aware, but it is conscious. A machine can be programmed to simulate intelligence, perform intelligent functions like playing chess better than some master, whereas a dog cannot, but the difference between the consciousness of a dog, or a human, regardless of capacity for self-awareness, and that of machine simulated intelligence, or rather, the distance between the two is infinity. They exist in difference realms, one of organic life, the other of inorganic matter.

There is NO linear progression between a machine's simulation of intelligence and consciousness. Why? Because the distance between the two is not finite, it is infinite. In other words, Moore's law won't get you to consciousness, ever. In my view, even the term 'intelligence' is not quite as accurate as the term consciousness. See, machines can simulate intelligence, but they cannot become conscious, because consciousness connotes awareness, self-awareness or not. A machine simply is not aware, as it is not life. Awareness is consciousness and consciousness is life and a machine is not, nor can it ever be, life.

A dog has awareness, it's just that it's not self-aware, or rather, it's not aware that it is aware, but it is aware--it is, therefore, conscious, nevertheless. And thus a machine can only simulate intelligence. In my view, AI, artificial intelligence, is the wrong term. The more accurate term should have been, 'simulated intelligence', or SI. The reason I say that is because the term 'Artificial Intelligence', if that is possible, then why not 'Artificial Consciousness'? It suggests that consciousness is achievable, i.e., it can become 'alive'. No, it can't. This is why I feel that simulated intelligence is the better phrase.

Now, if you want argue that intelligence and consciousness are the same thing, fine. I'm establishing here, for this argument, they are not precisely the same, though they are related. Intelligence is observable via functionality, capacity to perform complex tasks, where consciousness is only recognizable in the sense of animal or human has self will, has emotions, is spirited, is a rather unpredictable sentient being capable of free will, whereas a machine, that aspect can only be simulated. The distance between simulated life and actual life is infinite.

I personally would like to take it to a spiritual level, that consciousness, i.e., 'life' has a spiritual basis, ( which is why a machine can never reach it) but most scientists would not be comfortable doing that, as it is not falsifiable ( not now, anyway) so I will leave it out of the discussion.

Now then, I'm not saying mankind cannot manipulate natural organic processes, and produce some kind of organic humanoid which becomes conscious via a natural process, as mankind has created hybrid animals, so it may create test tube born humanoid creatures of the future, but this is not machinery, this is organics -- here, man isn't actually 'manufacturing' consciousness, he is manipulating natural processes, similar to how man created corn. Nor, am I saying that inorganic matter cannot interface with organic matter, and, in fact, that has been proven.

 

My belief: consciousness is all there is. It is the entire universe. Certain entities, for example the human species, can resonate better with this fundamental consciousness of the universe, than, say, a tree or a fish.

A.I. exists at a very superficial chemical level and will only ever have extremely limited consciousness, like a rock.

 

While the OP makes good points I suspect that there is a fundamental programing issue that needs to be solved in order to achieve AI or SI if the OP prefers.

Awareness is predicated upon the ability to grasp abstract concepts like I think therefore I am since it self awareness that is required.
Computers can be programmed to be aware of their surroundings by giving them sensors similar to our own that detect visible, audible, touch, taste and smell stimuli.

What is subsequently programmed when these stimulations occur is what matters.

Let us use the example of an audible question asking "Are you aware?" of a binary computer. The answer in binary is yes or no and therefore limited to what is programmed as to being the correct answer.

So what happens when the same question is asked of a quantum computer where the answer can be yes, no and other. The programmer now can use the other option to explore the facets of awareness that would comprise the level of awareness of the program itself. The programmer can even program self learning so as to raise the level of awareness based upon extending the knowledge base of what constitutes awareness.

Theoretically AI is possible. What it will take to achieve it is still TBD.

 

AI is SI and it is NOT 'consciousness'. It's machine versus life. consciousness is life, and whatever AI has, it's not life.

To simulate life is not to be life. The distance between the two are infinite. There is no linear progression from a machine to a living thing.

 

The definition of awareness is whether or not you would be able to differentiate if you were dealing with a machine or a person online. If you cannot distinguish between the two then that means that AI/SI has reached the point where it has reached the same level.

 

No, it just means you were fooled by a machine. It does not give the machine life or self awareness. A bird may be fooled by a birdcall but it doesn't give awareness to the birdcall.

 

Yes, it must be in the organic component that brings life. And life comes from life. We can make some impressive machines but not one with natural irritability. It is neat to see an interface between hardware and software however.

 

Well, simulated intelligence running on classic computers are probably not self-aware, but I don't recall any tests used to certify that premise.

There was a poster on here years ago who stated that in the quantum field there have been tests that consciousness does collapse the wave-function. And it's not only humans that can do this. Even a fly can collapse a wave function.

Or to be more specific, if a conscious mind views the internal systems of a quantum computer then that quantum computer stops working, both humans and flys do it.

Now, if we want to test that AI is conscious then we'd have to see if an AI viewing the internals of a quantum computer collapses the wave-function and causes the quantum computer to stop working. But, it costs millions of dollars for access to a quantum computer and it would really get expensive to test, which is probably why no one thought to do it yet.

 

NO! Just because you are fooled by a machine doesn't mean it's 'alive'. In fact, the distance between a machine that fools you and being alive is infinite. Self awareness means being alive, it's consciousness. Machines do not, nor will never have, consciousness.

 

Let's remember that humans have implemented consciousness and self awareness by means of a meat computer that is fully contained inside our bodies.

It is NOT infinite.

So, I don't see anything claimed so far to be an indication that creating this capability is impossible.

It is certainly WAY beyond our current technology. There is no doubt about that. And, some of the dimensions seem seriously daunting for technology to conquer. For example, speed, size and power consumption are all dimensions where our meat computers beat to death anything we have made from silicon. And, I think these probably would need to be solved. For example, if what we create takes more power than the sun or takes a century to decide not to hate itself, I'd say it's still no more than an indication of possibility.

Beyond that, the reference to AI is really just a hint that maybe it isn't necessary to slavishly model our meat computer in silicon (or whatever other technology).

Our brains are one solution - not a proof that there is no other solution.

 

We are discussing AI/SI and all it takes for awareness is to be aware of oneself. It is possible to program that awareness into the system just like anything else.

There is no actual requirement for there to be a biological component as far as intelligence is concerned.

Intelligence is the ability to learn and apply the learned knowledge.

Awareness is a facet of knowledge based upon perception of stimuli.

Someone in a coma is alive but unaware and incapable of any intelligent ability to learn and apply knowledge.

A machine that can learn from stimuli and apply the learned knowledge in a manner that demonstrates self awareness has AI/SI.

To put this in simple terms if the machine has control of a robotic armature and learns that using the off switch will shut it down and then makes the decision to use or not use the switch based upon that knowledge is applying awareness of self.

 

That machine does not have 'free will'. It requires programming. That programming might simulate free will, but simulation is not life. In short, that machine is merely an extension of a human programmer.

Self awareness is consciousness, consciousness is life. Machines are not 'alive'.

Well, I'll accept that you are fooled by the machine, that's about it.

You are conflating machine artificial intelligence with consciousness.

They are light years apart. In fact, the distance is infinite. There is no linear progression from one to the other.

Sorry, there is no evidence a machine is 'alive', has sentient feelings, etc., nor is there evidence a machine is aware that it is aware, for such can only be simulated. The mistake of logic you are making is the assumption that if it can fool a human being, it is 'alive'. It most certainly is not. You are confusing a simulation with a genuine living thing.

 

Nope, that is NOT what I am doing. I am making the DISTINCTION perfectly clear.

I am using the factual definitions and applying them EQUALLY to both biological and machine entities when it comes to intelligence and awareness.

A newborn is alive but not self aware and needs to LEARN self awareness through stimuli.

The same principle applies to a AI/SI.

Learning can be programmed and that means that it is technically possible for a machine to reach self awareness.

Just because we have not yet achieved this level of programming does not mean that it is impossible.

As WRM mentioned above we do not know what it will take and whether it will be worth the cost but it is theoretically feasible and we have already witnessed a revolution in information technology that has spanned the world in what was essentially just 4 decades.

That information revolution has raised our own awareness of how interconnected we are in both physical and knowledge terms about ourselves as a species and our impact on our environment.

The potential for AI/SI exists.

Making it into a reality is the challenge but we put a man on the moon and have discovered planets around other stars and connected the world so as to make this discussion possible even though none of us are in the same place so why is the AI/SI challenge beyond our abilities?

 

As with all technology, it isn't possible until it is.

I certainly won't be volunteering to have my consciousness transferred into a machine, no matter how lifelike the result appears (because we all know somebody else eventually will).

 

@Patricio Da Silva , you are equating consciousness of self to life. While our current technology is not capable of producing it without life, I wouldn't say it is impossible to get to it, a few centuries of technological evolution ahead. Why is life a pre-requisite to consciousness of self?

There is the Turing Test - can a computer pass for a human? Sure, at this point in our technology, only if the computer fools us, like it was mentioned above.

But I wouldn't exclude that at some point in a few centuries, if our species doesn't self-destroy before that, a Turing Test would be successful without being merely a simulation. Why not?

---------------

Have you watched the movie Ex-Machina? I found it truly fascinating. One of the best I've ever watched.

 

We know the distance is not infinite, because it is fully contained within a finite space.

Let's remember that when we start out, it takes years of interaction for our brains to acquire the information needed to become independent.

Humans come with a starter set. We have hardware specifically designed to learn language - how to store words in our meat dictionary, how to deal with verbs vs adjectives. How to do grammar, etc., etc., etc. (Certainly not limited to language.) We get a lot of starter set.

Any machine built to have human brain capability could rightfully expect to come with a similar starter set and to require years of learning from constant human interaction.

Again, "alive" means something else. To be considered living, something has to have each of seven characteristics - movement, breathing or respiration, excretion, growth, sensitivity and reproduction. An AI would not have to have all those things to be equivalent to a human brain in terms of factors such as self awareness, consciousness, free will, etc. If an AI can't breathe, get bigger, reproduce or poop, that's really not an issue.


As for "fooled by the machine", you're referring to the Turing Test. I do not even SLIGHTLY believe that's the hard part of creating a human brain. I think being truly conscious of self is a far larger issue.

Here's an article about Eugene Goostman, a 13 year old boy AI that fooled 22 judges.

https://www.wildfirepr.com/blog/can...signed,20th-century mathematician Alan Turing.

Two points raised in the article:
- could YOU pass that test? Would your answers convince a panel of 22 judges that you are NOT an AI?

- that article is entirely truthful. But, did you notice that it was fully researched and written by an AI?

 

Self awareness is life. It took nature some four billion years to endow organic matter with life, and another hundreds of millions of years to reach a level where an organism was self aware after the organisms became conscious.

This idea that man can endow a machine with life, which took nature eons, is the zenith of arrogance and reflects a fundamental failure to grasp what life, what consciousness, ergo 'self awareness' actually is. I repeat, the distance between actual life and the simulation of life is infinite in that the former is alive, and the latter is dead. There is NO linear distance progression between the two.

 

A child becomes self aware because of growth, which occurs concomitantly with stimuli.

A child, at any stage of awareness, is conscious. A machine can simulate such things, but does not possess that which a child possesses, which is life.

The difference between organic life and inorganic simulated life is that the former is alive, and the latter is dead. The distance between the two is infinite. A machine cannot, via progression, reach life. It will never become 'alive'. Mankind cannot endow matter with life, only nature can do that, though it does it with organic matter. Man can manipulate, edit, organic matter, but man is not the source of life, only nature does that, and the precise means. method, mechanism nature achieves it is a mystery.

 

To say that a machine fools people and therefore equals life is a logical fallacy.

 

No one except you is making that strawman argument.

Let's go back to the newborn infant example.

PRIOR to achieving "self awareness" the infant learns awareness of it's surroundings and gains motor skills to interact with it's environment.

Machines are reaching this level of motor skills and awareness in order to interact with their environment.

Self driving vehicles have not yet reached the sophistication to be 100% autonomous but they are aware of their environment and how to interact with it in order to navigate from one place to another. That did NOT take "billions" or even "millions" of years although one could argue that there have been millions of man years in the development of the automobile to reach this stage of development however in elapsed time it is little more than a century since the first motorized carriages appeared on our roads.

Probably in the next decade or so autonomous vehicles will become commercially viable because there are BILLIONS of DOLLARS to be gained by putting them on the roads. They will be the "toddler" stage of machines on the path to self awareness.

And since you raised the point about ARROGANCE it is the height of arrogance to ASSUME that ONLY carbon based life is capable of awareness and intelligence.

There is the distinct possibility of silicon based life forms existing in our universe and it is even possible that machine intelligence and self awareness could come about through the use of silicon bonds to form complex memory and processing functions in the future.

ARTIFICIAL intelligence is just what it says it is, artificial.

It does NOT require any carbon biology to achieve self awareness. Technically it is FEASIBLE.

 

Just to add to the interesting comments I suspect AI will not really take off until AI can autonomously program itself.

I don't think anyone can really predict with the outcome would be and I'm not sure if that would make it alive or conscious. It is the desire for greater evolution and happiness, whether base or spiritual, that seems to make animals and humans come alive. What is the 'life' purpose of a machine?

 

The ANSWER is 42!

Sorry, could not resist!

Machine AI will probably have as varied an array of the "life purposes" as ourselves.

The most basic life purposes are survival and procreation but we are not at this level when it comes to AI.

This is more akin to what are the life purposes of intelligent beings?

We have achieved the basic life purposes in abundance and now we seek other purposes in our lives.

Those range from the arts to the sciences and everything in between with some striving for excellence in each area.

Machine AI will be given a focus such as to deliver X from point A to point B or to grow Y without harming the environment or any other challenge that requires the machine intelligence to solve problems that we deem necessary.

 

AI isn't magic. It's just a piece of software analizing a set of data and branching its execution according to a set of predetermine rules. It doesn't even understand the meaning of the data or its branching. It only follow the rules set up by its programmer.

 

One would need to define what self-awareness means. IMO one definition is self-awareness is the ability to adjust behaviour in different circumstances for a net gain. Currently AI does not do that. I would argue that a dog is more self-aware than a human baby - a dog understands the difference between right and wrong behaviour and adjusts its behaviour when being watched. Is anyone fully "self aware"? That lack of self-awareness is evident when we dream ( third of our life). There is no reason to believe that AI will not eventually become just as self-aware as any human

 

But aren't brains doing the same thing fundamentally?