Hello folks!
I'm interested in your opinion about AI (artificial intelligence). What do you think when will be "born" and how will AI react to our world?
Printable View
Hello folks!
I'm interested in your opinion about AI (artificial intelligence). What do you think when will be "born" and how will AI react to our world?
Hmm...I don't quite understand your question, but I assume you mean when will they be humanlike. Well, let me tell you.
Robots, AI, and all that kind of stuff will never be like the movie A.I. To be like that, a programmer has to program a robot to do all those things in his mind. The programmer would have to have super evil intentions. Therefore, he will be arrested and it will never happen...
I think the question should be what is intelligence?
Is it the ability to make a decision? Most decisions we make are based on knowledge, we dont have to work anything out.
When I viewed your post I searched my knowledge to see if I had anything to add(same way a computer searches a file).
When I found relevant data I posted a reply(same way a computer outputs data in a file to the screen). No intelligence in that I just followed a routine.
So to summerise I think we are all just machines who follow rules the same as computers
But what about emotions, and spontaneous actions? Were those just programmed into us? What about learning? Is our brain just a big hard drive? We aren't programmed. The reason computers are much like us, is because they were modeled after humans, because we seem so near perfection.
gnu-ehacks:
How can you say something like that??
AI will be someday as smart as we are and even more. The only question here is when.
And about emotions. You know, everything is in math, including emotions.
What are you talking about? Do you know anything about proramming AI stuff? Something might be able to emulate the expression of emotions, but the AI can never FEEL those emotions. As an example, if a person's mother dies, the person will feel loss. If a robot's creator dies, it might be able to create a look of sadness on their face, and maybe even emulate crying, but a robot will never FEEL true sorrow, like you, me, or any other person would feel.
why not?! Our brains are just some kind of formula. AI just needs a hi level of intelligence and computing power to deal with such a thing.
have you had any experience programming AI? You can say FeelSad() and it will carry out a few things, but it's just a computer program. Human life is much more complex than any computer program will ever be. Human thought is not just a group of subroutines. It is a much greater thing, that's why you feel so happy when you get a computer program working. A program can emulate these feelings outwardly, but it never really cares. A program just does work, it never gets tired.
Your imagination about AI is a little mistaken. Of course it's a program but it's a program with intelligence.
Explain to me how to implement this intelligence then...
I didn't say that I can implement or write the AI. I'm just saying that someday AI will be born and on some stage she will be more intelligent than we are (humans). I recommend you to read this site Kurzweil AI.
We had a discussion like this already, not long ago.
It boils down to one key issue:
If human behavior is deterministic (the behavior of an individual is governed ONLY by the laws of physics and chemistry, not by free will) then computers can be as good as humans.
Computers can never have free will; freedom to choose and the total determinism of an electronic system are not compatible.
IF you were to think about life in general and try to work it all into a formula, it would need to have so many various sub routines that it would be just too big to be of any use.
Image programming a human hand - we can emulate its movement and improve on it .. but we can't produce a program or a device to fully emulate its ability to feel and its ability to work even if the item it was holding was not quite what it thought it was e.g. a hot drink in a wine glass.
Now imagine having to program all this ability into a PC and THEN add the ability to imagine an image of the future and describe it. - totally impossible
- and we have yet to add the ability to have emotion -
AI - it really stands for All Imagination!
In Kurzweil's articles, I read some interesting things. It said that eventually a brain will be able to compute and hold more knowledge than a human. But again, this is nothing REAL. Do you know the reason why it is called AI? Because it is ARTIFICIAL. While it is possible to emulate human thinking, it is not possible to emulate human FEELING. To this point, a robot cannot even truly SEE the way we can. They can find depth by a form of echolocation, and they can see heat sources, but they cannot truly see and process this information the way a human can.
I tend to disagree; computers have already been shown to handle an incredible variety of stimuli.Quote:
Originally posted by f0ul
IF you were to think about life in general and try to work it all into a formula, it would need to have so many various sub routines that it would be just too big to be of any use.
Image programming a human hand - we can emulate its movement and improve on it .. but we can't produce a program or a device to fully emulate its ability to feel and its ability to work even if the item it was holding was not quite what it thought it was e.g. a hot drink in a wine glass.
Now imagine having to program all this ability into a PC and THEN add the ability to imagine an image of the future and describe it. - totally impossible
- and we have yet to add the ability to have emotion -
AI - it really stands for All Imagination!
The thing is, say you have a dozen variables of interest. You don't need to explicity define, beforehand, how the machine ought to behave for every possible value of each of these. You need to give it ENOUGH points, but not all of them.
A neural network can, if not easily, at least potentially map any function, in any number of variables, arbitrarily close. Such a solution EXISTS (finding it is rather hard, usually).
By making programs which are trained, the program can extrapolate from what it DOES know to what it does not. It's like a person - there are many different kinds of mechanical pencils, yet you can extrapolate from the kids you may have used, to learn how to use ones of a different style.
Already, there is an AI which can fly an F-15 as good as any military pilot. It's a complex system, but combinations of neural networks, possibly with fuzzy logic controllers, can learn from experience, the same way a person does.