Artificial Intelligence/Human Intelligence - Playing GOD?

[Benedikt]

Hi there,

I have a question about AI.
Scientists try to create computers with AI. That's okay IMHO. But here's my question:

To create computers with AI - to make computers more human-like isn't it?
But if you make a computer think and behave like a man, then you have all the mistakes of mankind also inside of the computer. I mean, if they try to make computers/machines "think and decide like humans are able to", ok then you have a device acting like a human - with all the pros and cons.

For example:
Computers are kinda "serial organized" machines, which are really capable of calculating large equation systems.... and so on. But today's computer's would never be able to, for example, cross a road with heavy traffic. You would need massively parallel systems for that kind of jobs... image recognition, quick decisions, a "feel" for reality, just like the human mind. But if you create a machine like a human mind, then it would easily pass such tests, and fail with large equation systems again!

So what do we want? Machines acting like human beings, or should I say machines acting like machines and a "human" part added?
With all the scientists trying to build machines that have personal experience, machines that are able to talk ... etc.. wouln't it be better to say instead of
ARTIFICIAL INTELLIGENCE -> HUMAN-LIKE INTELLIGENCE?

And BTW, what is intelligent? The way humans act is intelligent, really? A human decides with his/her own knowledge what's intelligent and what's not? What would extraterrestrial beings say about - the "human intelligence"?

So many questions, so much to discuss... Feel free to reply to my post!
Maybe you have some links for me?

Greetings and I'm looking forward to your requests!

Bene

 

[dejitaru]

What is god?
AI is man-made, so it can only be as good as the man who made it. There is yet to be a program written that equals a human brain.
If there were the perfect android, it would still be an android.

 

[Shalmanese]

AI researchers have practically given up on "human-like" AI and are currently working on "human augmenting" AI. That is, AI that works in areas where humans perform poorly. Rather than trying to create any overarching intelligence, they usually try to create a very specific type of intelligence.

 

[grant2]

Originally posted by: dejitaru
AI is man-made, so it can only be as good as the man who made it.

How do you figure that?

A car is man-made, yet it is faster than the (men) who made it.

A crane is man-made, yet it can lift much more than the (men) who made it.

A computer is man-made, yet it can perform complex calculations much faster than the (men) who made it.

 

[Evadman]

Originally posted by: grant2

Originally posted by: dejitaru
AI is man-made, so it can only be as good as the man who made it.

How do you figure that?

A car is man-made, yet it is faster than the (men) who made it.

A crane is man-made, yet it can lift much more than the (men) who made it.

A computer is man-made, yet it can perform complex calculations much faster than the (men) who made it.

But all of them requre man to pilot them in one way or another. ( or a woman ) The gar that pickups groceries for you is still a ways off. Expecialy one that figures out what you need.

 

[everman]

I think the ultimate goal would be a system that could design better versions of itself in a manner far superior to that of any man. That would mean no longer any need for humans to make these systems. I wouldn't count on that any time very soon though
Current AI is more about taking a specific task or set of tasks and designing a system that can perform it better than humans can.

 

here is a question for you...suppose we do create AI...would it be ethical to destroy or shut down the computer that has the AI, as it would in essance be killing an intelligent being.

 

[dejitaru]

here is a question for you...suppose we do create AI...would it be ethical to destroy or shut down the computer that has the AI, as it would in essance be killing an intelligent being.

Nope. But humans and their "feelings" could be tricked into thinking it would be wrong.
It's just data loss, and still, you could create another "organism" exactly like it (unlike if you killed a person).
Though, as a tech, I would be opposed to destroying such a refined program.

 

[FuriousBroccoli]

Dejitaru, You say 'Nope' so bluntly as if you understood the nature of sentience. Not knowing what sentinence is means you don't know what it isn't, so don't be so quick to jump to conclusions. It's stupid to rely on your basic human intuition here, because we're well beyond that. We're computers too, just made out of cells. Do you think there's something magical about us that couldn't exist in a silicon based life-form?

 

[dejitaru]

Do you think there's something magical about us that couldn't exist in a silicon based life-form?

no

What are you saying?

A computer program is property, not life.

 

[Benedikt]

Do you think there's something magical about us that couldn't exist in a silicon based life-form?

Yes, the soul.
And the little mistakes that make us human, but if you find a way to create them also in a silicon based life form?

Greetings,

BW

 

[Benedikt]

What about massively parallel organized tasks to do.... like image recognition (I mentioned it before)?
Any ways to find a "human mind simulation"?

Greetings

BW

 

[FuriousBroccoli]

Originally posted by: Benedikt

Do you think there's something magical about us that couldn't exist in a silicon based life-form?

Yes, the soul.
And the little mistakes that make us human, but if you find a way to create them also in a silicon based life form?

Attributing sentience to a soul is like saying physics is the work of god -- it's an utterly useless concept in advancing our understanding of the universe, and you will never become any wiser because of it.

To answer your original question, you have to look at the goals of A.I. Simulating the human mind is an extremely small, and scientificially fruitless (for the time being) part of the whole field of A.I. Right now A.I. is used to solve problems that can be modelled as inputs and outputs to systems that, at the most basic levels, could be considered forms of artificial intelligence, in the asbtract scientific sense.

For example, you mention image recognition. We don't need HUMAN A.I. systems, we just need A.I. systems. These are nothing alike, and you need to do some reading to appreciate the vast differences.

The only reason to ever make a computer like a human is for the purposes of our emotional needs. For example, an A.I. child or lover ;-)

BTW, a computer could easily be taught to cross the road, I could write a program to do that myself, in under a day. Of course it would be using sensors not a camera. Using a camera would take a few weeks, but it's still relatively easy.

Computers *will* become more intelligent than us, and be able to do everything we do. It's not sci-fi, it's science, the only argument is over how long it will take.

 

[FuriousBroccoli]

Originally posted by: dejitaru

Do you think there's something magical about us that couldn't exist in a silicon based life-form?

no

What are you saying?

A computer program is property, not life.

What on earth do the two have to do with each other? Can you give one reason you think computers couldn't be considered sentient? One real reason that is. For example, saying that they are merely bunches of transistors isn't a real reason, because we are merely bunches of molecules.

 

[Benedikt]

I wouldn't say the soul is NOT important for , or NOT important for science.
It's just the ---uh- let's say, the FLAIR of a human - the feel - the impression a human leaves to his/her environment.... hmm quite hard to say.

But if I would be an AI creator, I would say it's definitely important for AI, at least if you try to create a lover/boy like you mentioned... A machine with a soul? Hmm...

And I must admit my street-crossing example is a bad one.
Let's say playing an instrument, doing good painting or craftsmanship, or having the feeling as a film director to make good movies....
You can make computers play instruments, but how? There's so much between ....

Greetings

BW

 

[FuriousBroccoli]

You have to look at this from a more reductionist perspective. Everything that seems amazing or magical about us can be broken down into simpler functions that computers can emulate. Not yet of course, but in time, perhaps even centuries or longer, whatever behaviour or vibe a real person gives you that makes you think they have a soul or are somehow beyond science, can and will be replicated by an andriod.

 

[dejitaru]

Originally posted by: FuriousBroccoli
You have to look at this from a more reductionist perspective. Everything that seems amazing or magical about us can be broken down into simpler functions that computers can emulate. Not yet of course, but in time, perhaps even centuries or longer, whatever behaviour or vibe a real person gives you that makes you think they have a soul or are somehow beyond science, can and will be replicated by an andriod.

A computer only does what you tell it to do.
If you believe that to be life, fine. I don't.
There are no images, thoughts, or feelings, only numbers hacked to incorporate error. Cold steel in a nice package. They don't think, they calculate.
The human brain can study itself, yet not quite understand it. If you create the perfect android, its race would eventually die by some unseen or neglected flaw.

A machine can emulate a human, but it could not be human.

I could write a program to emulate my tarantulae perfectly. Would it be as good as a real spider? No. It would not be able to feed itself or reproduce.
Even the most well-written virus would rely on its users to run it, as all software.
A program can only execute very specific functions. Even if those functions are many, it's still a robot.

Do you consider a virus to be living? A brick? Your PC?

 

[soltrain]

Everything that seems amazing or magical about us can be broken down into simpler functions that computers can emulate.

Reminds me of some quote or other.. "Any sufficiently advanced technology is indistinguishable from magic"

 

[Agent004]

A computer only does what you tell it to do.
If you believe that to be life, fine. I don't.
There are no images, thoughts, or feelings, only numbers hacked to incorporate error. Cold steel in a nice package. They don't think, they calculate.
The human brain can study itself, yet not quite understand it. If you create the perfect android, its race would eventually die by some unseen or neglected flaw.

You sure? If you don't know anything, you will also do as you are told(ie, when you are a young child, you simply don't understand why. you just do it)

It can be also consider humans has no such thing as feelings as such, since they are just eletrical pulses and how we intepret them is similar to how AI would repond to data or codes. Human is in the same sense, collection of molecules/body tissues

If you programme the AI to study, and expose things to it...... eventuay it will also study itself. Plus, it's not about issue what the consequence of a perfect android is, it's about whether it's possibility to create such android

 

[dejitaru]

You sure? If you don't know anything, you will also do as you are told(ie, when you are a young child, you simply don't understand why. you just do it)

Yet if your computer didn't know anything, it couldn't learn. It would remain still without its program.
If told by someone their own age, a small child is less likely to follow blindly. There is consideration and judgment.

It can be also consider humans has no such thing as feelings as such, since they are just eletrical pulses and how we intepret them is similar to how AI would repond to data or codes. Human is in the same sense, collection of molecules/body tissues

As anything else. These are only incorporated into the program as interpretations of their symptoms. There is no suppression.
Emotions don't make life.

Plus, it's not about issue what the consequence of a perfect android is, it's about whether it's possibility to create such android

Yeah, but a team of humans over years couldn't do it all. They could emulate the practical functions, plenty to fool anyone.
Yet complexity isn't the issue.
The simplest organism will always be greater than the best robot.

 

[jackwhitter]

when scientists attempt to create a human AI, i do not think they are trying to create something that is like humans on a day to day basis... i think they are trying to create something that questions its own existance. something that will better itself outside the bounds of it's programming. a machine that is self-aware more than one that can mimic some human abilities of quick decision-making based on possibilies built into it's programming. even if a machine can "learn" while running and change/add/remove possible outcomes, the ones we have right now are not self-aware or consious. i think that is what they mean by human AI. a machine that is self aware will be able to read people's emotions (their body reactions to a specific circumstance), and place their emotions in relation to itself (the machine). if someone is repulsed by the attempt at creating a conscious being, then the AI would interpret this (although, only through a series of observations of temperature variences, body movements, etc and then it has to process these observations through a set of parameters (read program) and determine what they mean.) a computer could be made that could do this today, BUT a human would have to input all the possibilites of the human race's variance. the self-aware AI would be able to adjust it's perspective as it aged/time passed.

our self-conciousness plays a large role in our emotions/humanity. without being aware of ourself (and caring for that matter), then everything is a cold observation. some people today are accused of being cold and machine like... the machine that recognizes itself as an entity that reacts and is reacted to will be the closest to human AI.

 

[FuriousBroccoli]

Dejitaru, you aren't thinking ahead enough, your basing your ideas off our current technology, not future technology. You are assuming we would create a passive database type program to emulate humans. Scientists know full well that we couldn't begin to emulate humans using such rudimentary techniques. Well we can do a lot better than this. Given enough time, we can replicate the kind of neural networks humans have, so that a computer will be more than a passive program, it will be an entity which can be taught, can make decisions, and can be self-aware. When that happens, *all* your points are void, except one -- Is it conscious?

 

[dejitaru]

They are rendered.

Perhaps new laws would arise regarding murder of these golems.

Built from chips or cells, it would still have not a father, yet a programmer.

Are humans conscious? What is life? We're just programmed from birth to think we are.
Self-awareness promotes survival, but it's all fake. Of course it would think, just like us.
You live, reproduce, die. But for what?
It would be self-aware, because you told it to be.

Would this creation have a soul? Do humans have souls?

It isn't free will, yet a program.
An android can defy its intended instruction and kill its creator. (like in the sci-fi movies)
A human can defy its instruction and kill itself.

It will be quite a while before these golems are considered life by the masses, scientists, or the law.

Humans are just the invention of some race who's forgotten about us.