Design of artificial intelligence must read

Design of artificial intelligence must read

[1 paradox]Why 0.999... is not equal to 1?

Written in 2012

The current mathematic theory tells us, 1>0.9, 1>0.99, 1>0.999, ..., but at last it says 1=0.999..., a negation of itself (Proof 0.999... =1: 1/9=0.111..., 1/9x9=1, 0.111...x9=0.999..., so 1=0.999...). So it is totally a paradox, name it as 【1 paradox】. You see this is a mathematic problem at first, actually it is a philosophic problem. Then we can resolve it. Because math is a incomplete theory, only philosophy could be a complete one. The answer is that 0.999... is not equal to 1. Because of these reasons:

1. The infinite world and finite world.

We live in one world but made up of two parts: the infinite part and the finite part. But we develop our mathematic system based on the finite part, because we never entered into the infinite part. Your attention, God is in it.

0.999... is a number in the infinite world, but 1 is a number in the finite world. For example, 1 represents an apple. But then 0.999...? We don't know. That is to say, we can't use a number in the infinite world to plus a number in the finite world. For example, an apple plus an apple, we say it is 1+1=2, we get two apples, but if it is an apple plus a banana, we only can say we get two fruits. The key problem is we don't know what is 0.999..., we can get nothing. So we can't say 9+0.999...=9.999... or 10, etc.

We can use "infinite world" and "finite world" to resolve some of zeno's paradox, too.

2. lim0.999...=1, not 0.999...=1.

3.The indeterminate principle.

Because of the indeterminate principle, 1/9 is not equal to 0.111....

For example, cut an apple into nine equal parts, then every part of it is 1/9. But if you use different measure tools to measure the volume of every part, it is indeterminate. That is to say, you may find the volume could not exactly be 0.111..., but it would be 0.123, 0.1142, or 0.11425, etc.

Now we end a biggest mathematical crisis. But most important is this standpoint tells us, our world is only a sample from a sample space. When you realized this, and that the current probability theory is wrong, when you find the Meta-sample-space, you would be able to create a real AI-system. It will indicate that there must be one God-system in the system, which is the controller. Look our world, there must be one God, as for us, only some robots. Maybe we are in a God's game, WHO KNOWS?

相关阅读
★概率论：完全可能性的理论与现实图景
★悖论问题的统一解
★从延安文艺座谈会到人类社会的未来
★发现号航天飞机——自然算法伟大的飞矢变换
★星际争霸1的AI设计思路：以人族开局为例
★诸神之战在星际争霸1的实现[001]AI游戏的发端

More info, two other download points(written in Chinese):
(1)speedyshare.com/DQz9y/AiforSC.rar
(2)filerio.in/kw4cl2l2y3qi

 

Log in or Sign up to hide all adverts.

That's nothing but a waste of space. Next time, try posting something something of actual value. <shrug>

 

Log in or Sign up to hide all adverts.

i find it of immense value.
just think of it:
bank manager: you have overdrawn to the tune of 50 trillion dollars.
you: i'm sorry sir, some chinese dude said that 1.999=2, them pennies sure do add up quick don't they?

 

Log in or Sign up to hide all adverts.

Yes I agree, it is extremely relevent! It is similar to a program posted some years ago which counts from zero to one going half the distance each time. 'x' eventually reaches one but it does not stop immediately upon reaching one! I believe the problerm is the values become infinitely small. The program is as follows:

x=0
y=1

while x<y
{

y=y/2
x=x+y

}

 

I find some of its concepts very difficult to understand and some are very ill logical.

 

I think this thread was started by artificial intelligence.

Why does 1= 0.999 unless we are rounding up for sake of simplicity.

How does any of this math relate to AI?

 

0.999 is not equal to 1, but 0.999... (recurring) is. If you want proof, consider the fact that three thirds of something constitutes the whole, even though one third of 10, for example, is 3.333... In other words, 9.999... = 10

Your calculator will agree.

Beats me.

 

You don't need this kind of accuracy for creative thinking.

The mind is actually very inexact, but it is intuitive. IMO, this is the problem with AI. How to make it inductive and intuitive.

And probably the greatest obstacle is to fashion an artificial "mirror neural network" which allows the AI to "recognize" and respond to stimuli experienced by others.

 

"True" AI can never be accomplished unless we can figure out how to give it the ability of original thinking. Until then, it will be limited by it's programming and cannot "think outside the box."

As a case in point, does anyone remember how Kekulé's, who's most famous work was on the structure of benzene came up with the solution? He dreamed of a snake swallowing it's tail. So just how would we go about endowing a computer with that kind of ability?

 

@ read only,

Even if a computer that had a neural network and learned without programming could exist, there is likely no viable purpose for it. A computer that does not follow a program is not easily controlled. It would be an engineering marvel and possibly the creation of some sort of life, but nobody would fund such research.

It is nice to say we want a computer that can think creatively, but nobody wants an accounting program that does math like a human, or a car gps system that decides to alter the destination on a whim.

In some advanced civilization perhaps garage mechanics research may unite into some semblance of a mind that can create music and art, but I don't see it as likely for hundreds of years if then.

 

In order for that to happen, someone is going to have to do some serious redefining of what "life" is. For example, 'self-reproducing' will have to be removed.

 

In the Book by James Glieck called "Chaos Theory", there is a piece on how Lorentz (Known for things other than just the Lorentz transformation) spotted during a weather pattern predication algorithm that the program ran completely differently when the same figures were entered. On further investigation it turned out the problem that caused the immense entropic difference was just down to a difference in how many levels of decimal placing was used in each run, the first run a colleague had "rounded" figures up.

This is one of the basis's of fractals, which itself can be infinite or eventually become finite repeating patterns.

As for "1's and 0's", anyone dealing with artificial intelligence development to my knowledge doesn't limit themselves to a Binary(Integer) definition of prospective results, most have likely been taught about "Fuzzy Logic"[wikipedia], In fact "Fuzzy Logic" is one of the main points and problems that occurs with various Database related projects.

An example of this, in the UK a company called ATOS has been employed by the state to create a database on whether people are employable or classed as incapacitated or disabled, the problem is for the most part ATOS has built it's database with "Boolean" values to work out how a person is, when in fact a person being an "Anologue system" comprises of more data subsets than their databases can handle. This in turn has meant that ATOS has misread and misrepresented hundreds of thousands of people, potentially placing volatile or at risk people back into the workplace.

Fuzzy Logic is Heuristically dealt with through the process of Neural Networking.

 

A computer based AI will never understand what a number is. If it can be called reasoning, they must reason with inputs that ultimately convert everything to binary numbers.

A human will never understand what a length actually is. Galileo did not. Newton did not; Einstein did not; yet it's mostly what these, our greatest minds, ever dealt with. We reason with senses that provide neural sensory inputs that are analogs of length for vision, hearing, touch, taste, and smell. Infinity makes no more sense to a finite machine simulation of a mind than an infinite length does to our own manifestly finite minds and limited senses.

AI camps that deal with symbology are determined to create an artificial intelligence that overcomes some of the human mind's greatest limitations and weaknesses, such as believing in and being immersed in lies. An inordinate amount of our brainpower is associated with modeling social behaviors of others. Who needs a machine to do something as useless as that? Machines already exist that can create their own languages and communicate, but for the most part these are no better optimized than most human languages.

AI camps that are behaviorists are more likely to produce machine intelligences similar to our own. These machines will optimize the use relatively simple senses to explore their world and obtain needed resources for self-organization and cooperation with humans. Motivational systems like our own endorphins will be needed to optimize performance that accomplish needed goals.

Does any of this help?

 

To create a.i. i beleive power is the biggest problem: if you create artificial life it must never be switched off (unless done so by itself.) I have tried discovering the solution to mather's halving program but after many pages of long division i concluded that a solution will never be reached. A computer is simply rounding.

I beleive any kind of a.i. will be purely mathematical: computers were originally for numerical calculation, and even now are simply CALCULATORS! I think geeks with calculators are the future (if i am permitted to post such a thing. ;o) )

 

It occurs to me that we take the human brain for granted, as a flawless computer. This is far from true. Our brain is able only to observe and compare input which is perceived by our senses. But our senses and computing powers are extremely limited in the grander scope of universal functions.

This can easily be demonstrated by optical illusions which produce a "false" images in our mind.

Moreover many other species possess the ability to perceive touch, smell, hearing, and sight much more precise and in a much greater range than humans. This is the reason why we have need for instruments to perceive and measure environmental input outside the scope of our natural ability.

But in the end AI functions the same as our brain. The problem lies in achieving flexibility of information association at deeper levels than just processing information as initially perceived.

I wonder if a computer would be fooled by this little example:

Please Register or Log in to view the hidden image!

 

I doubt a computer would be distracted by that picture. Experiments have shown so. However a.i. must be aware of other humans! I would be very interested in helping you develop any a.i. programs you have. Of course i couldnt do it for you because then it wouldnt be your program, but as is taught in software development, you must have an idea first and the code follows.

 

I think this site should impliment computer code that erases any thread started by someone with there first post and where there isn't a follow-up post within 24 hours

Please Register or Log in to view the hidden image!