Can AI eventually become 'self-aware'?

.......

 

Mental capacity is not something I would even consider when a question of conciousness comes up. It is whether the creature can feel anything. Or if it can see anything. I would venture to say a camera sees nothing. It is just an advanced light box.

 

OK. I just don't see that as related to the issue of the capability of the human brain.

I do believe we will at some point have the ability to create life, too. But, I'm not so sure how meaningful that is. We can already engineer life forms to work for us in various surprising ways - creating needed compounds, eating plastics and oil, attacking cancer cells (still only a possibility?), delivering medicine to the needed location, etc.

I'm definitely in favor of not modifying humans. And, I'm concerned that it will be exceptionally hard to resist as we learn to make "improvements" to human embryos.

 

Yes. I see all those as mental capacity - that is, it is the brain that is making the difference.

One might think of sensory elements, facial expression, arm waving, noise, language, light collection, etc., as part of the I/O technology.

But, I think consciousness, self awareness, our emotional capacities are more than that, too. Our human understanding of ourselves differentiates us from most other life forms.

 

You need to explain why you see that as an issue.

 

I don't exclude that machine intelligence may one day become self-aware. You say it's all programming... well, maybe in the future, a quantum AI or something, will write its own lines of code for its next thoughts or moves. You call it fake, it's just a judgment of value. If machine AI gets to be self-aware, it's not "fake." It's just self-awareness achieved by different means than what our neurons do. Anyway, I don't see much point in continuing to debate this point. You seem to be set in your ways and not willing to listen to a different opinion, so we should just agree to disagree.

 

I bring the OP because I've seen too many times people suggesting that man can endow life into inorganic matter whereupon they do not understand that 'self awareness is consciousness, and consciousness is life. A machine can simulate, but it's not the real deal.

My view is that that is impossible.

 

Self awareness is life. Man cannot endow life, only nature can do that. Man can engineer genetics, but that's not 'endowment' that's steering a natural process, in the same way man has made corn from maize.

 

Certainly, AI can become conscious. Consciousness is nothing more than simple self-awareness and awareness of any type can be programmed into a computer

And computers will never become evil and turn against us. Unless, of course, we program them to

 

When you think about it we are programmed individuals. So I don't know if AI can become self aware. But two things to consider..... could we? And should we?

 

I think that for now we can't, but in the future we may be able to. As for should we, I don't really buy into the apocalypse that the machines will take over. They'd still have many limitations. See, the dolphins seem to be self-aware, and they haven't taken over.

 

That's the Turing Test and although Turing was brilliant, I don't think he was able to predict the sophisticated machine learning that would allow programs to learn from interactions with a human. Otherwise chatbots would be considered "alive."

So I don't think the Turing Test is a valid way to determine self awareness.

 

Your still conflating these two issues. I think they are separate.

If some future effort duplicates the full function of the human brain, I don't really see what being "alive" has to offer.

I'd point out that simulation is an approach to creating a solution. In software, engineers sometimes use that approach to creating a solution.

Whether that is the strategy that will finally be successful is a question that probably can't be answered. But, there certainly is a lot of effort to simulate collections of artificial "brain cells" that work like real brain cells work.

 

But that's the premise, that 'self awareness' is consciousness, and consciousness is life, and that machines are NOT life, no matter how strongly they are convincing.

 

I haven't seen someone besides you who suggests "consciousness is life", "requires life", etc. I don't mean to limit your claim here - I don't want to play word games.

Biology does't agree with that. Biology has criteria for what life is. Consciousness is not included.

Given that we have no artificial consciousness at present, it's certainly the case that all examples of consciousness that exist today come from living forms. But, that doesn't make it a rule or requirement.

 

Biology has one requirement and that is irritability. In this case "feeling" should be defined as an act of an organism reacting to a stimulus. In some way it is conscious of its environment. What would you call a sentient being if not alive? Even a being that doesn't fullfill all the requirements for biological life if it is sentient it is alive.... right?

 

The Turing test was NEVER considered to address self awareness.

 

I don't believe you have proposed a justification for suggesting that being alive is a requirement for consciousness.

It's true that human brains are part of a living entity. But, that's not evidence that being part of a living entity is an absolute requirement.

How about a thought experiment? Brain cells are reasonably simple in terms of how they function. We have working models of brain cells, including small numbers of brain cells that interact as a group. Now, let's say we create 10 billion of those cells - the number of brain cells humans have. Then, let's say we connect them in the way that our brain connects them.

I would say that sounds HORRIFYINGLY hard - certainly totally impossible today.

But, I don't see anything in that which would require being a life form.

And, that is only one route for building a brain. It could well be that we need to take other routes, using quantum computing, or some other as yet undeveloped technology.

 

Turing knew about learning.

Chatbots aren't alive, because they don't poop, reproduce, or otherwise have the 7 or so characteristics that biology defines as requirements to be considered living.

 

Prehaps "alive" is the wrong term. Maybe self aware, or intelligent would fit better.

 

I'm not trying to be stubborn here, but I do think there is a real definition of "life" today. And, I don't believe that definiion involves consciousness or applies to artifically constructed machines.

I believe people who walk into a room where there is a biological human and an artificial human are not going to consider them both as "alive". And, that will be true for biologists.

Maybe you could argue for some other terminology.

We may well need new words - especially since we have no idea of what such new devices might consist of.

 

I like that terminology suggestion!

I think today's chatbots can't pass the Turing test (that is, fool a human who is focused on determining if it is human), but at some point it may well be the case that we have artificial "brains" that are evenself aware/conscious. There do exist machines that have passed that test, but I think they are mostly from places doing serious research on the topic.

There are science museums that have bots that wander the facility and talk to guests. They tend to be NO challenge to the Turing test, but they can be disarming when they recognize you based only on voice or "vision" when they run into you again and continue past discussion.


I'd suggest we are a LONG way from self awareness/consciousness. I ONLY argue that it can't be written off as impossible - mainly because we have an example (a human brain) that is not infinite in any dimension.

Who knows - we may develop laws preventing that from happening! There is certainly a body of sci fi related to artificial intelligences that have a serously unfortnate level of self awareness and free will.

 

Furthermore none of those 7 characteristics of life include self awareness, conscience or intelligence which is the bogus premise that the OP is promoting.

Intelligence can be defined to include reasoning, problem solving, decision making, the ability to learn and adaptability to a changing environment.

All of those characteristics can be programmed to some degree with our current technology. It is a matter of developing those programs and introducing additional aspects of intelligence like abstract thought and conceiving new possibilities that are currently beyond what we can program. Those aspects might require biological components but that is yet TBD.

 

I believe I will be proven correct, in time.

 

Laws eh? Maybe three of them should suffice?

But I think a chatbot has passed the Turing Test.

A Chatbot Has 'Passed' The Turing Test For The First Time

But that's different from actually being self aware. It only needs to fool humans into thinking it is to pass. For may humans, that's a pretty low bar.