Can Computers Think?

[Protector one]

Originally Posted by toyoka

Bottom line; Computers are only as smart as their creator. So the computer's perception of thought is more or less predetermined.

No, that's not necessarily true. Systems with the ability to learn can become 'smarter' (strictly regarding a specific domain, usually) than their creators.
Heck, they don't even need to be able to learn. They just have to be more efficient in some way. Calculators can perform calculations faster than their creators can. (Cheap example, but a good one! :P)

[Siberianhusky89]

Originally Posted by Protector one

No, that's not necessarily true. Systems with the ability to learn can become 'smarter' (strictly regarding a specific domain, usually) than their creators.
Heck, they don't even need to be able to learn. They just have to be more efficient in some way. Calculators can perform calculations faster than their creators can. (Cheap example, but a good one! :P)

I'm not sure about the calculator thing... Aren't the numbers programmed into the calculator? Plus, computers don't understand the concept or existance of a number, it just displays something. Kind of like if you writer 2+2 on a paper, and cover the answer up with another piece of paper and as if there was a set list for equations. Plus, if I remember correctly, a computer can't learn anything outside of programming, though, perhaps there might be a way to do so, then again, I could be wrong. Anyhow, if you were to go up to a robot that didn't know how to jump and tried to jump in front of it so that it could possibly try it's self, chances are that the robot is never going to learn even by analyzing you.

[toyoka]

Unless it's Asimo It can probably jump!

[Protector one]

Originally Posted by Siberianhusky89

I'm not sure about the calculator thing... Aren't the numbers programmed into the calculator?

Even if the numbers were just pre-programmed (which they're not), you could see this as a form of intelligence.
Thinking about this example reminds me of the discussion around Searle's Chinese Room. Again, the notion of intelligence and thinking is very subjective; this is something that becomes painfully clear when a group of people are confronted by Searle's Chinese Room.
(For the record, I think the 'system as a whole'—that is: the room, the man, the input, the instructions and the man's work—can be considered to 'understand Chinese', even if the man itself does not.)

Originally Posted by Siberianhusky89

Plus, if I remember correctly, a computer can't learn anything outside of programming

Well, that's not true. Learning systems can acquire new knowledge through data. This data may be fed to it, it might gather it itself (for example, from the internet), or it might create data itself through exploration.
You might want to look into 'Machine Learning' once. It's a wonderful field!

[Siberianhusky89]

Originally Posted by Protector one

Even if the numbers were just pre-programmed (which they're not), you could see this as a form of intelligence.
Thinking about this example reminds me of the discussion around Searle's Chinese Room. Again, the notion of intelligence and thinking is very subjective; this is something that becomes painfully clear when a group of people are confronted by Searle's Chinese Room.
(For the record, I think the 'system as a whole'—that is: the room, the man, the input, the instructions and the man's work—can be considered to 'understand Chinese', even if the man itself does not.)



Well, that's not true. Learning systems can acquire new knowledge through data. This data may be fed to it, it might gather it itself (for example, from the internet), or it might create data itself through exploration.
You might want to look into 'Machine Learning' once. It's a wonderful field!

First part is interesting however, the second part is a bit off. A computer CAN gain data off the internet, however, that data was programmed to do so, therefore, programming is pretty much almost, if not completely, involved and was programmed to be input into the computer. I've learned that computers can only understand the concept of "on" and "off" (A.K.A the Binary Language 1's and 0's). But if that's not the case, then what is?

[Nagaoka]

Frankly the way I see it our brains basically work like very advanced computers. We think in a certain way because our brain has been wired that way. Alter the way our brain is wired and we think in a different way. If AI advances far enough then yeah I can see a computer having thoughts.

[Protector one]

Originally Posted by Siberianhusky89

A computer CAN gain data off the internet, however, that data was programmed to do so

No, not necessarily. For example, a program that is intended to learn natural languages could use any piece of written text (or even spoken text) on the internet for its purposes. A NL-learner could be using this very forum post to improve (or deteriorate) its POS-tagging accuracy as we speek!

[Claxon]

Originally Posted by Nagaoka

Frankly the way I see it our brains basically work like very advanced computers. We think in a certain way because our brain has been wired that way. Alter the way our brain is wired and we think in a different way. If AI advances far enough then yeah I can see a computer having thoughts.

Yes this is much the way I think of it. Computers transmit signals through electricity to determine what they should do, human brains & nerves transmit signals using chemical reactions. Yes they can only do things that they've been programmed to do (including being programmed to learn new things), but who here instinctively knew how to read and write without being taught? It's a fun little subject

[WeTheTwisted]

I once attempted to code a simple chat-bot in C#.

This made me think a lot about human factors which make each person an individual, morals and such.

It really is, all invisible variables. So if all the neccesary ones are replicated in code and the computer used them as we use ours, like "Hmm, Tom just broke up with his girlfriend, I should not mention that I have met a nice lady...".

Well, that's a lovely example. That defines who we are:

Nice person - "Hmm, Tom just broke up with his girlfriend, I should not mention that I have met a nice lady..."

Bad person - "HA! Tom's missus ditched him, i'm totally gonna rub it in his face that I met a nice lady".

That would be tact I guess and morals... maybe?
Either way, variables!

Then if you just coded one of those "20 Question" information network bots to it, it would have a means to learn about many things:

"A duck quacks."

That would tell it that an object called a "Duck" performs an action which is a "quack". It would probably already know what a duck is (animal, vegetable or mineral route).

So if you could tell it to parse data from Wikipedia or something instead of being fed information by many users... surely there's a means you could replicate a human.

This train of thought could literally go on for hours...

[Siberianhusky89]

Originally Posted by WeTheTwisted

I once attempted to code a simple chat-bot in C#.

This made me think a lot about human factors which make each person an individual, morals and such.

It really is, all invisible variables. So if all the neccesary ones are replicated in code and the computer used them as we use ours, like "Hmm, Tom just broke up with his girlfriend, I should not mention that I have met a nice lady...".

Well, that's a lovely example. That defines who we are:

Nice person - "Hmm, Tom just broke up with his girlfriend, I should not mention that I have met a nice lady..."

Bad person - "HA! Tom's missus ditched him, i'm totally gonna rub it in his face that I met a nice lady".

That would be tact I guess and morals... maybe?
Either way, variables!

Then if you just coded one of those "20 Question" information network bots to it, it would have a means to learn about many things:

"A duck quacks."

That would tell it that an object called a "Duck" performs an action which is a "quack". It would probably already know what a duck is (animal, vegetable or mineral route).

So if you could tell it to parse data from Wikipedia or something instead of being fed information by many users... surely there's a means you could replicate a human.

This train of thought could literally go on for hours...

I believe that something like that deals more with Value than with morals, however, I do get what you're saying. Though, morals are so controversial, and computers would only be able to "know" the morals that someone programmed into it. Example: Divine Command - A belief that whatever God commands is morally right or wrong. Of course, God might not even exist for all we know. There also exists other theories such as Ethical Egoism. You're Ethical if you think about yourself and yourself only. If a computer were to be programmed with this thought process, it would be useless. Why do I say that? Because the purpose of us making computers and robots is to help us in our everyday life.

@Protector One

Interesting, however, I've read that computers can't comprehend the Words that are displayed on a screen, they just display them and such, so they don't actually understand English, German, etc... Or were you saying something else that I may have misinterpreted? "^^