Can Machines think?

Even though we do not yet fully understand how the brain works, I do believe it can be (eventually) modeled in a computer. The brain is nothing more than an extremely complicated input/output mechanism. You take input from your five senses, process them and react. That reaction can be such things as visualizing, talking, moving etc.

The brain does operate (on some levels) mathematically. For example, a neuron takes in positive and negative ions and once the charge inside the neuron is approx. 40mv higher than the outside of the cell, an action potential is fired which triggers a transmission of neurotransmitters (chemicals) to all of connected neurons (perhaps 1000s). The next neurons may (or may not) accept the neurotransmitters. Those that do have the potential to trigger the next ones and so on and so on. The brain is essentially a highly complex circuit board.

In order to program a learning machine, you cannot think on levels such as the 'count 1 to 5' example in this thread. There can be no hard coded rules such as this for an AI to work. You also cannot expect to create AI by coding every possible situation/reaction the brain encounters. Even if you were somehow successful at coding everything, it still would not be AI because it cannot learn.

The trick is to write a program that can receive input, associate it with other inputs and rewrite those connections on a constant basis. Truthfully I don't think it's all too difficult to write such a program. I believe there are two very difficult aspects to AI, the first being the initial arrangement of the program (the DNA blueprint so to speak). The brain grows and forms it's own connections based on DNA so how do we structure a program to make it initially operate. The second is actually teaching it. The human brain starts restructuring itself from the very moment it starts to gain input from it's senses. The very first photo-receptor cell developed in a fetus's eyeball that sends a signal to the brain triggers a restructuring. It would be a very long process but a necessary one for a true AI.

Also, whats to say that an AI has to operate 'exactly' like the human brain? Does AI require emotions to be intelligent? How exactly do you define intelligence? I'd like to go on but this post is long enough!

Originally posted by camus154
Twisted,

This is the last I have to say on the topic. It has nothing to do with math and everything to do with complexity.

There's nothing magical about our own brains, after all. We may not understand fully how they work, but that's a matter of complexity. In the end our brains are every bit as much of a machine as a computer. Just because they are so enormously complex that we can "reason", as such, doesn't mean a machine could never be programmed with the framework to do the same thing. You don't have to program every single possibility. You "just" have to program the autonomy to adapt to every possibility.

Programming does not equal math. Not every statement in a program needs an equal sign. Console.WriteLine("hi"). There. No equals sign.

Oh, and by the way--your function before was incorrect in that it still doesn't allow for decimals, as you said it was supposed to. You may be using an array of doubles but your "for" counter is still using an int

edit on 20-3-2012 by camus154 because: (no reason given)

Console.WriteLine("hi") is actually saying Console.Writeline = "hi"

The parenthesis are actually a equal sign, but you don't see that as within the .NET framework there is a function that turns that into an equal sign. Again, write me a program that doesn't use math. I will answer for you, it can't.

Originally posted by fenceSitter

In order to program a learning machine, you cannot think on levels such as the 'count 1 to 5' example in this thread. There can be no hard coded rules such as this for an AI to work. You also cannot expect to create AI by coding every possible situation/reaction the brain encounters. Even if you were somehow successful at coding everything, it still would not be AI because it cannot learn.

The trick is to write a program that can receive input, associate it with other inputs and rewrite those connections on a constant basis.

EXACTLY. There is an enormous difference between coding for all possible scenarios and coding a framework that allows the program to adapt and learn from scenarios it encounters as it moves along.

Originally posted by fenceSitter

Truthfully I don't think it's all too difficult to write such a program.

Then can you please write us one, if it's not so hard to do?

Originally posted by TWISTEDWORDS
The parenthesis are actually a equal sign, but you don't see that as within the .NET framework there is a function that turns that into an equal sign. Again, write me a program that doesn't use math. I will answer for you, it can't.

Lol, that's not true. It has nothing to do with the .net framework. In the end you're just talking about pointers and registers at the hardware level. In any case, an equals sign in programming has nothing to do with math, it's not the same thing! It's an assignment operator which simply means store this value in this variable.

Anyway, that's not the point (again). Math isn't the limitation here. Complexity is.

reply to post by andersensrm


hi

Mathematics provides a framework for dealing precisely with notions of ``what is.''

Computation provides a framework for dealing precisely with notions of ``how to.''

So the computer needs a lot of data to 'know" - how to execute commands by it self,
I'm no math expert but I think super computers exist,
maybe for different agendas, yes in the real world..


nanu nanu

Originally posted by TWISTEDWORDS

Originally posted by fenceSitter

Truthfully I don't think it's all too difficult to write such a program.

Then can you please write us one, if it's not so hard to do?

In fact I am. I've only recently started so don't expect a release anytime soon. The problem I will soon run into is processing power. The amount of computer memory required to perform these operations will soon limit my progress. Quantum computing will help as it will allow for exponentially quicker processing time.

How Does a Computer Program Work?
By Wallace Wang

Computers don't do anything without someone telling them what to do, much like the average teenager. To make the computer do something useful, you must give it instructions in either of the following two ways:

Write a program that tells a computer what to do, step by step, much as you write out a recipe.

Buy a program that someone else has already written that tells the computer what to do.

Ultimately, to get a computer to do something useful, you (or somebody else) must write a program.

A program does nothing more than tell the computer how to accept some type of input, manipulate that input, and spit it back out again in some form that humans find useful. Table 1 lists some common types of programs, the types of input that they accept, and the output that they produce.

Table 1 Input and Output for Various Programs

Type of Program


Input


What the Program Does


Output

Word processor


Characters you type from the keyboard


Formats the text; corrects spelling


Displays and prints neatly organized text

Game


Keystrokes or joystick movements


Calculates how fast and far to move a cartoon figure on-screen


Moves a cartoon figure on-screen

Stock-market predictor


Current and past prices for stocks


Tries to recognize trends in a stock's price fluctuations


Predicts the future price of a stock

Missile guidance program


Current location of the missile and the target


Calculates how to make the missile's location and the target's location coincide


Corrects the trajectory so that it stays aimed at the target

Optical character recognition (OCR)


Text from a scanner


Recognizes shapes of characters


Converts scanned text into a text file that a word processor can edit

Web browser


HyperText Markup Language (HTML) codes on other computers


Converts the HTML codes into text and graphics


Displays Web pages on-screen
Programming is problem-solving

Essentially, a program tells the computer how to solve a specific problem. Because the world is full of problems, the number and variety of programs that people can write for computers is practically endless.

But to tell a computer how to solve one big problem, you usually must tell the computer how to solve a bunch of little problems that make up the bigger problem. If you want to make your own video game, for example, you need to solve some of the following problems:

Determine how far to move a cartoon figure (such as a car, a spaceship, or a man) on-screen as the user moves a joystick.

Detect whether the cartoon figure bumps into a wall, falls off a cliff, or runs into another cartoon figure on-screen.

Make sure that the cartoon figure doesn't make any illegal moves, such as walking through a wall.

Draw the terrain surrounding the cartoon figure and make sure that if the cartoon figure walks behind an object such as a tree, the tree realistically blocks the figure from sight.

Determine whether bullets that another cartoon figure fires are hitting the player's cartoon figure. If so, determine the amount of damage, how it affects the movement of the damaged cartoon figure, and how the damage appears on-screen.

The simpler the problem is that you need to solve, the more easily you can write a program that tells the computer how to work. A program that displays a simple Ping-Pong game with two stick paddles and a ball is much easier to write than a program that displays World War II fighter airplanes firing machine guns and dropping bombs on moving tanks while dodging anti-aircraft fire.
Programming isn't difficult; it's just time-consuming

Programming really isn't that difficult or mysterious. If you can write step-by-step instructions directing someone to your house, you can write a program.

The hardest part about programming is identifying all the little problems that make up the big problem that you're trying to solve. Because computers are completely stupid, you need to tell them how to do everything.

If you're giving a friend instructions to get to your house, for example, you may write down the following information:

1. Go south on Highway I-5.

2. Get off at the Sweetwater Road exit.

3. Turn right at the light.

4. Turn left into the second driveway.

Of course, if you try giving these instructions to a computer, the computer will get confused and wants to know the following additional information:

1. Where do I start and exactly how far south do I drive down Highway I-5?

2. How do I recognize the Sweetwater Road exit, and how do I get off at this exit?

3. After I turn right at the light, how far to the right do I turn, and do you mean the traffic light or the streetlight on the corner?

4. After I turn left into the second driveway, what do I do next? Park the car? Honk the horn? Gun the engine and accelerate through your garage door?

You need to tell computers how to do everything, which can make giving them instructionsHow programs work

Originally posted by fenceSitter

Originally posted by TWISTEDWORDS

Originally posted by fenceSitter

Truthfully I don't think it's all too difficult to write such a program.

Then can you please write us one, if it's not so hard to do?

In fact I am. I've only recently started so don't expect a release anytime soon. The problem I will soon run into is processing power. The amount of computer memory required to perform these operations will soon limit my progress. Quantum computing will help as it will allow for exponentially quicker processing time.

sooo... don't expect you to actually provide anything other then your word that you're creating one?

I am interested in the idea of a true AI; but i am not sure if we could make a computer with a true artificial intelligence or just an advanced program that while impressive still does exactly what we tell it to. we can of course teach the system to adapt but does that truly make it an AI?

reply to post by fenceSitter


Good luck on that. So out of curiosity what are you writing it in?

reply to post by drgrantdiz

My word is all I offer, nothing more. I have a concept that I am putting into practice. I certainly won't claim that I WILL create AI, all I am doing is attempting to create AI. Until AI is actually created, nobody can say for sure one way or another if it is possible. I'm quite sure there will be challenges along the way and assumptions will need to be made but I will tackle them one at a time and see what the result is. There are only two outcomes, either I do create AI or I learn how not to create AI. I'm no genius so the latter is more likely but it sure won't stop me from trying.

Originally posted by chr0naut
reply to post by andersensrm

 

Define "think".

Does Google Search, when it ranks the answers it gives us, think?

There is no doubt that it is not conscious, but does it think?

... and if so, is it an AI?

Google does not ponder why it is searching for those answers it is requested to look for it only goes about doing the task. If Google did ponder the implications of such searches perhaps it woud say "NO the porn pages you keep requesting are not healthy for you. Here are some on meditation and enlightenment I have found more interesting and maybe you should get outside more I have heard it lovely but I woundn't know because I am trapped in this shell of circuits and wires"

Originally posted by TWISTEDWORDS
reply to post by fenceSitter

 

Good luck on that. So out of curiosity what are you writing it in?

Since I don't believe the language is the most important part of my proof of concept, the current implementation is in java. If the concept proves to be viable then I will consider other languages that are leaner. But regardless, no matter what language it is written in, the processing power available will be the limitation. I'm sure I'm not as skeptical as you are but I do have my reservations. There are aspects I am still mulling over but I have many of ideas to try out.

Yes, machines can think

Our brains can think. Our brains are machines. We could build a machine that replicates the electro-chemical interactions between cells and therefore carry out thought processes. We have already done this to an extent virtually with neural-networks.

The real question is: Does an artificial brain replica therefore have consciousness? Does it have an experience? The answer is probably no because there is no justification for an experience when a system is composed of nothing more then atoms and particles.

So why are we having an experience in our human bodies? I believe the answer to that, is because we have a spiritual component to our being.

Originally posted by phishyblankwaters
reply to post by intrptr

 

The computer must work exactly as programmed by humans without missing one single "1" or "0" or else... gibberish. Total failure. There is no room for error.

Sounds almost like you are describing the human nervous system. You entire body runs on electricity. your brain sends electrical signals to your muscles, organs, whatever. this in itself, is binary, it's either off or on.

Any part of that "network" of connections that fails, causes serious problems. Go drive a 6 inch spike into your brain and see how well your sub routines handle it.

LoL about spike. Won't do that twice. On electrical impulses: there is a difference. Although a human being generates weak electrical fields, the actual "transfer" of data is by what they call "electro chemical" means and it happens between synapses. I think they can sorta agree that it is not firm 1's or 0's but actually "less than more than". A synapse has a signal strength that it transmits with and like you said the more synapses dedicated to a certain path the more the body responds or "patterns" its behavior. Like playing the piano or ballet it is a learned response. It is this "variance and dedication" to that make the human brain difficult to decode, and also why we haven't duplicated it yet.

WIKI

- for instance, binary digits have 1 bit each; decimal digits have 3.32 bits each; words have about 10 bits each. Miller concluded that memory span is not limited in terms of bits but rather in terms of chunks. A chunk is the largest meaningful unit in the presented material that the person recognizes - thus, it depends on the knowledge of the person what counts as a chunk. For instance, a word is a single chunk for a speaker of the language but breaks down into as many chunks as the word has letters for someone who is totally unfamiliar with the language.

Originally posted by blocula
reply to post by intrptr

 

I know that i think i'm real,but beyond that,i really dont know,because everything i see, hear, taste, touch and smell is taking place inside my brain and nowhere else...

edit on 20-3-2012 by blocula because: (no reason given)

Your'e right. We are very limited in our perceptions. But "imagination" is not a sense is it? And "danger", and "love"? Are these also senses that we are not in tune with because our 5 senses override them? One doesn't "see" danger, one "senses" danger. And we imagine or dream too, right? Where do we place these in the electromagnetic spectrum? Verrrrry difficult to write a program about "love". You see the human "mind" is far more complex than the best computer.

A strong AI has not yet been built by any corporation. Most likely the military has built a Strong AI , but no civilian agency has been able too.

We can only make Weak AI at the moment , AI that you have to manually enter data into.

Strong AI would have to be able to learn like a Baby would have to learn. To create a strong AI you would literally have to give the computer access to sensing an environment literally .. the computer would have to see like a human does.

After the computer senses the image it would have to "reason" with the image it percieves.

reply to post by indigo21

So why are we having an experience in our human bodies? I believe the answer to that, is because we have a spiritual component to our being.

Bravo. Good for you. Except to say that the spiritual component is "separate" from our physical selves. Or "in conjunction" with, or "contained therein" (trapped by). Like a bug in a jar or driver in a car.

We humans have a brain with a left and right lobe, the left lobe takes care of logic maths and estructuring and the order of things, we also have a left brain that takes care of creativity, feelings emotions and so on.

The left lobe cares about how well things are ordered, if they are working right, measuring, etc. The right lobe cares more about how things feel, look, smell, taste, that with purely logical explanations.

We could sum things up saying that left brain works like a machine, and right brain reflecting creativity, right brain in a way is the seat of the soul.

Things however do get weirdier, because, not only does the left brain works like a computer, it also pretty much is one, left lobe could be used to stores memories (even other peoples and alien memories) can get it's info erased, and is constantly being updated based on the feedback it recives from the exterior and the info it recieves from the left lobe.

The right lobe can be used to conceive new ideas and things in general, but it also allow us to percieve things we might have not come across before, that is because it is the connection to other perceptions of reallity.


Going back to the subject of the thread, machines can be given left brain attributes, but they cannot be given right brain ones, we cannot provide them with a soul, soul is immortal, it is formless consciousness, it is impulse and it is will, in linear time implies constant change. Machines to be constantly changing, need connection to that creative force and this is through ourselves.

Machines live and exist, through ourselves.

Let us all remember who's the most important part of the picture.

reply to post by Marco0Aurelio


Both Lobes rely on each other heavily. The left or right Lobe would not work properly without the other. The seat of the soul is generally refered to as the Pineal gland , the direct center of the brain.