Staredit Network

Is it possible to make AI using computers?
I, being a programmer, know how programs work.
I think that if you had a program that stored every bit of information it found, then crossrefenced everything, and found similarities, it would develop into AI. That is provided you also give it Pleasure/Pain/Pressure receptors.

Some people think that computers simply can't cut it, which is what this poll is about. But there has recently been the invention of Neural Network, which are, virtually, Artificial Brains. Thusfar, Nueral Networks can learn little, and run out of space very, very fast, but they act almost identically to human brains. Even though, these Neural Networks show no emotion at all.

The farthest they have gotten, I beleive, is getting a Neural Network to control a ragdoll in a Physics Simulator, with accurate muscle movements and all. The Neural Network learned to walk correctly, but never figured out how to climb the stairs. They checked it and all its memory cells were completely filled. The Nueral Network, they theorize, could entirely make it up the stairs were it given more memory capacity.

I found this information about four years ago, and by now I'm willing to bet they have made much progress. Of course, if it did really well, you can bet the military has them working under top secret cover now.

It's very well possible. I can see a point where people won't really hate each other, they would hate the robots instead, cause the robots are stealing jobs and are considered a "Secret threat". Not a bad idea if you ask me, help unify people a little more. Really debateable argument, I havn't debated in it well enough to have a firm stance yet.

No. I'm also a programmer, and I know about NN (they really aren't new, either).

Everything in the program is, well, programmed. It will always do what the programmer says, without variation. If you want to call that intelligence, fine, but it's not.

You can make a program that wont listen to the user.
AI in starcraft, for example. Type 'Stop'.
Does it stop?

That doesn't mean anything Shapechanger. But I believe AI is possible.

Make a computer that is able to make choices based on the current situation and that is AI. It thinks and callculates the probabiilty of everything, but in the end it makes its own choice. Making choices is intelligent.

Well, that's not neccessarily true. The AI in CS:S, for example, will 'Choose' whether to camp or attack. That does not prove it's intelligent.

if by AI u mean something like what we think with, then obviously..
We were created, if it was by nature.
Which proves its biologically possible.
Why not technologically possible?
yes, with both.

I guess so. But when will AI become a huge part of our lives?

I still think AI does exist.
Behind top secret doors.

AI is like starcraft triggers, if the condition is to "Bring marine to Create Unit" then the action will by "Create 1 firebat". So if on AI if the condition is "apple is in front of you" then the action might be "pick up the apple" So ya... That's my take on it.

No, that's mindless following of if statements.
Nowadays, we use AI to describe computers that play games against us.
That is not the AI I mean.
I mean real, true Artificial Intelligence.

I think it is possible through computers, since they will probably be the main system of the AI, like our systems that allow us to survive. The only thing I'm unsure of is if we can control them if they begin to think.

Dude, AI rebelling is a Sci-Fi fantasy, elaborated through movies and books.
We would not be stupid enough to give AI access to nuclear weapons.

The Matrix.

Don't even go there.

All intelligence is is combining many simple functions to make a much larger complex one. AI is extremely possible, just look at how our brains work. Chemical and electrical signals is what creates your intelligence, it doesn't come from anywhere else.

Being able to combine multiple things like when you see an apple pick it up with, if you are hungry and have an apple in your hand then begin eating it (and of course all the complex functions that would be in those actions). That is what creates intelligence. Obviously it gets much more complex then that.

Crossreferencing.
I think if you made a good crossreference program, it could think for itself when combined with input, output, pain, pleasure, neutral, a database, and instinct.

I was chatting to the creator of NERO last month and he agreed with me when I said it would be at least 50 years before any truly intelligent AI was developed.

NERO, incidentally, is an AI sandbox which you can download from the link above. You get little robots which you try to get to evolve into effective soldiers - try it.

well it seems possible to create a massive computer capable of storing/crossreferencing all data... unfortunetly it would need a massive storage compacting system and a massive data-base i'm guessing it couldnt store it in standard binary it would probably have to work at least greater than in base 1024 to even be able to begin to catagorize the data of our world... not only that but if it did it would obsorb data at a massive rate giving it acess to almost everything in our society... i extremely doubt this would lead to anything catastrophic like nuclear war or the "terminator" theory... but it would probably lead to technilogical advancments upon studying the "artificial intelligence" it all depends on how the AI acts once its obtained all the data of this world will it hunger for more or simply be happy monitering our cursed society (such as overseeing building construction, keeping track of what is in everything, what does what, who does what) it truely depends on how the ultimate intelligence would act so i really cant say... that also has a huge factor such as if it was simply a gatherer of data it would attempt to gather all data... yet if u programmed things such as pleasure or pain it would stay away from anything that had to do with those subjects and attempt to eithor find some substitute for the data or destroy the data all together... but if you programmed it to aquire data based on pre-created and alterable directives it would find all data and simply do nothing else... i truly cannot tell you what an AI will do or not do and thats what causes fear and doubt in society concerning computer intelligence... computers we use now have an IQ of zero and 100% they can learn anything and yet they do not learn it as it is simply stored and can be re-referenced

but back on topic there would really be no need to fear an AI system so i'm sure someone will attemt to make one and suceed... and if you fear an AI well then... you probably will not make one... it all comes down to the creators opinion on the subject if one will come to be or not come to be

You could just as easily say that the laws of reality are concrete, and that the entire future is already defined, there is no such thing as freewill, and there's nothing we can do about it.

I don't know why you would say that, though.



Also, about the poll: It makes no sense having a "computer" option and a "neural networks" option, considering NNs are run on computers.

This doesn't make any sense.

I don't know why you would say

though.

This is a very complicated question. Its possible, but not yet at the very least. You see, computers cannot come up with completly random numbers. They come up with suedo random numbers which are so random, we could never really tell the difference, but until we can find a way that it can actualy come up with a number 100% randomly, complete artifical intellegence is not possible.

So the question is, will we ever be able to make it come up with a 100% random number? Well, that is a question which obvisouly the answer so far is no to. Computer follow steps in order. Thats the way we programmed them. We would literaly have to recreate the computer if we wanted AI.

Well, my statement was true, but yours was false. That why I was confused.