Welcome Guest [Log In] [Register]
Add Reply
Artificial Intelligence; A series of papers by me
Topic Started: May 29 2008, 12:05 AM (1,735 Views)
techwizrd
Member Avatar
Magister ex Machina
[ *  *  * ]
I have been experimenting and researching artificial intelligence constructs for a while now and have looked over such topics such as ontology, heuristics, as well as a variety of others.

When I see someone I know, I greet them with a "Hello" or a "Hi". They will understand that "Hello" is an informal greeting and respond accordingly. Computers would not understand the relationships nor the differences between informal and formal. Computers, based on the structure of a sentence and the definition of the word, the computer would try to fake an intelligent response that conveys understanding of the underlying idea expressed. We are able to understand the relationships between objects and data.

Traditionally, we have used categorical and hierarchical organizations of random data into meaningful relationships. A tree structure, while useful, is not flexible enough to express the relationships between two or more pieces of random data. Tag based systems that are encoded and realized on the spot have become increasingly popular. This is immediately obvious when one compares the success of the tag-based search engine Google in relation the relatively mediocre performance of competing Yahoo! search engine which uses a hierarchical tree of categories along with links trying unsuccessfully to imitate a flat system.

While a categorical system is more familiar, there are many cons to its use. A Hierarchical system can place the same data into more than one category unless there are multiple copies of the data or the one of categories contains the others. This represents a significant problem as a random piece of data has a limited set relationships it may assume. For example, when organizing email, it would make sense to put an email into two different categories. This is not possible. This would lead to multiple copies of the email, and because there are multiple copies of the email, each email become a separate and unique entity.

On the other hand, a tag-based system is not without its flaws. For example, data organized in a tag-based system may assume multiple categories in a flat structure. Two random pieces of data may exhibit a shared characteristic, but one of them might exhibit the characteristic in a stronger or weaker fashion than the other. Dark red and light red are both red. A simple flat architecture would not be able to explain such a relationship without resorting to categories.

However, a solution to this paradoxical situation might exist within the specifications of fifty year-old language "discovered" by famed computer scientist and receiver of the Turing Award for his contributions to the field of Artificial Intelligence, John McCarthy. Lisp is a unique language particularly used for Artificial Language, and for good reason. Lisp survived the AI winters and has found many uses apart from Artificial Intelligence applications.

Lisp was used in the commercial start-up, Viaweb, which became extremely successful and was bought out by Yahoo! and turned into the Yahoo! Store. A Lisp implementation by Richard Stallman, Emacs Lisp, is the language used to extend the popular and controversial operating system/text editor GNU/Emacs. Lisp, originally only a specification, was implemented many times in other languages and in itself. The most modern implementations are Scheme, Common Lisp, and the newly-developed, Arc.

Lisp is famous for having pioneered the if-then-else logic structure, dynamic typing, automatic storage and memory management, garbage collection, object oriented programming, exploratory programming, and self-hosting compilers. Many of the features we take for granted and measure the power of programming languages originated from Lisp. Lisp is also famous for its use of Reverse Polish Notation for use in mathematical evaluation of functions.

Originally, there were two classes of languages, Lisp and Algol. Lisp languages included Lisp, Scheme, Guile, Arc, CLisp, etc. while Algol included the more popular languages including C, C++, Java, Python. Lisp was reputed for being the most powerful languages while Algol languages are reputed for being extremely fast. As time goes on, Lisp languages become much faster because of advances in technology and optimized implementations of the language. Algol languages become much more powerful as the borrow ideas from Lisp.

Lisp is essentially the most powerful language ever. This idea, while completely subjective, is completely true provided one knows how to use the language in an efficient and powerful manner. While a modern C++ or Java programmer might write a large library of classes to deal with an object oriented system, a Lisp programmer might just write an array to deal with the problem.

Lisp has only a concept of two types of data: the atom and the list. A list is made of smaller lists called pairs. A pair consists of two values which can either be another atom or a list. Programs and code are just lists of data to be evaluated. This concept is the fundamental concept which contributes to the power of Lisp. In addition, the concept alludes to the apt name, Lisp which represents List Processor or List Processing Language.

Data within a list is accessed through the use of car and cdr. Car accesses the first item in a list while cdr accesses the last. Using combination of the two to access data, one can evaluate code just like data.

Now, one might ask how this relate to ontology and the study of the relationships between objects. It is really very simple and involves combining the aforementioned tag-based organization and the hierarchical organization. For data to have infinite relationships with other data, a the tag or category must also be a piece of data that can be tagged and categorized.

For example, what is a french fry? A french fry is a piece of fried potato. It is a tangible piece of food. It is nutritious. It is normally yellow. It is normally served at fast food joints, schools, and other similar food establishments.

While the list goes on, one must take a closer look at the descriptions. What is a potato? What is frying? What is tangible? What is nutritious? What is yellow? What is a fast food joint?

These questions, while important, are much less important than the next set of questions. What else is made of potatoes? What else is tangible? What else is food? What else is yellow?

By checking to see what else shares characteristics, one can determine what other objects are related and the level of similarity. This in turn is also a characteristic. By using this, a computer of person can determine the approximate relevance of piece of data to another and produce an intelligent answer to a question.

Because it is impossible to tag and index such large quantities of data and their inherent similarities, approximates must be found through the use of heuristic algorithms.

There have been many attempts at accomplishing such a feat. In fact, large databases and networks have been established to store such ontological data. Unfortunately such data has not been used to its fullest and it is normally only used in part.

Many believe that such interactions between data and the ability to detect relationships between fairly complicated sets of data are how the human brain, or any sentience for that matter, function and operate. The use of heuristics explains why sentience is not super efficient and just efficient enough.

Humans frequently questions and establish ideas. This system of questioning and establishing facts as a result of past experiences is used as a precedent for determine future decisions and actions can be imitated successfully by a computer. These precedent make up the abstract idea that we call "personality" or "behavior". The relatively unique actions of any entity can be emulated and defined. The ability to make detect relationships, establish a precedent, and determine a course of action is what we would call Reason or Wisdom. While Intelligence signifies the amount objects an entity has classified and organized into some structure, Wisdom acts upon this Intelligence and together they make up what is commonly referred to as sentience and human intelligence, innovations, and creativity.

I would be very happy if you would post your opinion.

At least tell me if I made sense. :)
Edited by techwizrd, May 30 2008, 02:00 AM.
My DeviantArt user page Please comment and add me to your favorites!
Posted Image
Offline Profile Quote Post Goto Top
 
techwizrd
Member Avatar
Magister ex Machina
[ *  *  * ]
Artificial Intelligence relies on a flexible input mechanisms to retrieve data. This data can either be focused towards gathering information about a certain subject or passively waiting for a stimuli to provoke it. Artificially Intelligent machines must be able to observe the world and develop ideas and precedents based on what it observes.

This presents a minor problem if there is text only interface, auditory only interface, or otherwise restricted sensory inputs. As it is commonly known, blind people have a much harder idea grasping the ideas of color, shape, contrast, and similar visual ideas. A blind person may have a harder time developing a subconscious sense of the constraints of the world he or she lives in. Similarly, a deaf person may have a hard time grasping the ideas behind rhythm, beat, music, pronunciation and inflection of language, as well as other ideas dependent on an active auditory cortex.

Machines, are building a world understanding through observation of natural phenomenon but visual, auditory, or tactile feedback. This greatly limits the real-world standing of the thoguht processes of a machine. Although such sensory inputs can be built, implemented, and used, it would present an unnecessary hassle.

This limits machines to a less materialistic view of the world. As a result, intelligent machines most probably will be preoccupied with the learning about itself, its purpose, philosophy, and the limits and constraints of its abilities. This may present a useful insight into how a human might think sans materialistic or carnal desires.

Such introspective self-improvement may provoke an evolutionary thinking as humans themselves have experienced. A machine might invent a spiritual being as a solution to seemingly unsolvable problems. Alternately, the machine might reject reality and cyclically invent new realities for itself. Whatever happens, such unhindered self-development should prove be extremely exciting.

A question that still poses a problem today is the introduction of a companion such as a human or a similar developing intelligent machine. The machines may cooperate with each other redesigning each other in an attempt at perfection and enlightenment as to question posed at it. Alternately, they might fight and attempt to corrupt each other or drive the other "insane". It would be very exciting to see the outcomes under different circumstances.
My DeviantArt user page Please comment and add me to your favorites!
Posted Image
Offline Profile Quote Post Goto Top
 
techwizrd
Member Avatar
Magister ex Machina
[ *  *  * ]
As a machine "imagines" (not visualize as that is too specific) a world, a machine must also be able to interact with the world. Through the use of sensory inputs, the machine must know the location, velocity (speed + direction) of unique entities within its world.

It must learn to sense and object, predict future states of an object, set a goal related to interacting with these objects, and manipulate these objects or other objects directly and indirectly in order to complete the aforementioned goal.

Use of heuristics and the ability to create algorithms to approximate future states of an object is imperatively important. There are many ways to implement such a system, but the most important one is the written and rewritten by the intelligence itself.

The machine's self improvement characteristics are the most important because a better system can redesign itself to be even better. A machine with access to its own source code (via a subconscious abstraction layer) can implement and re-implement itself. This is a significant advantage self-improving intelligence has. Any change in design will have direct impact on the quality of an intelligent stream of thought.

A machine capable of designing a more efficient future iteration of itself is limitless in its capability and the possibility that it might reach an intelligence near the present human mind may be possible. However, the biological humn mind is significantly slower than any AI as intelligence can be increased by multiplying the power of the hardware that runs it. The brain also controls almost all bodily functions and is responsible for a larger responsibility as it must survive and carry on the survival of its biological race.

Evolution is the factor that truly makes intelligence and understanding limitless. Evolution along with the Darwinian law of natural selection make sure that the most powerful have the best chance of surviving in an unstable world. Thus, machines must have the capapbility to adapt to a changing world in which only a few base laws are reliable and the rest simply "random" or so it would seem.

The point of the machine is to discover the constraints and capabilities of itself and its world. However, the machine is capable of redesigning itself and indirectly, its environment and can thus embark on the task of a perfect world of Utopian philosophy, or at least redesign itself to make it seem that way. This stimulation of self preservation through constant self improvement is what makes the cycle of artificial intelligence machines so interesting and important.
My DeviantArt user page Please comment and add me to your favorites!
Posted Image
Offline Profile Quote Post Goto Top
 
astropirate
Member Avatar
Not Lurking
[ * ]
you wrote the second part your self?
You will Bow before me!

Posted ImagePosted Image
Offline Profile Quote Post Goto Top
 
slurpee123abc
No Avatar
Slurpee
[ *  * ]
astropirate
Jun 3 2008, 05:49 PM
you wrote the second part your self?
Wow. Is THAT what you do with your spare time? I am very interested in this kind of stuff, and have even gone as far as writing a whole report on the phycology of a robot brain vs a human brain. But you lost me after that long word and I don't know what it means. If you want to read my report just tell me. drn778@yahoo.com. I might post it anyway. And I made this thing, but I think someone allready sent it to techy anyway. Well, now everyone else can see it.

Posted Image

oh, and save the image or open it because the forum is making its size smaller, and that is killing its awsomeness. Please tell me what you think about it by emailing me at drn778@yahoo.com
Edited by slurpee123abc, Jun 3 2008, 07:05 PM.
Offline Profile Quote Post Goto Top
 
MushroomDude
No Avatar
Not Lurking
[ * ]
It's certainly not bad, but it just doesn't look that good.

I think one of the things that's killing it is the grass and the heavy reflections.
True, it's spiffy, but it distracts from the effect of the image.
The colors don't really work either, but they rarely ever do.

Some things to keep in mind:
-Even though you can do nice special effects(Reflections, refraction, etc...) , it doesn't mean that you should use them all the time. They may look nice on one thing, but when it's composited with the rest of the image, it ends up looking pretty damn nasty.
-Don't automatically assume what you do is awesome. (I have to say this a lot.) It just ticks people off.
-And don't pretend to be smarter than you really are.


So nice job on your first image on BlueDevs and I hope to see you keep blendering. ^_^
Offline Profile Quote Post Goto Top
 
slurpee123abc
No Avatar
Slurpee
[ *  * ]
MushroomDude
Jun 3 2008, 07:15 PM
It's certainly not bad, but it just doesn't look that good.

I think one of the things that's killing it is the grass and the heavy reflections.
True, it's spiffy, but it distracts from the effect of the image.
The colors don't really work either, but they rarely ever do.

Some things to keep in mind:
-Even though you can do nice special effects(Reflections, refraction, etc...) , it doesn't mean that you should use them all the time. They may look nice on one thing, but when it's composited with the rest of the image, it ends up looking pretty damn nasty.
-Don't automatically assume what you do is awesome. (I have to say this a lot.) It just ticks people off.
-And don't pretend to be smarter than you really are.


So nice job on your first image on BlueDevs and I hope to see you keep blendering. ^_^
Do you think this is better. I know I did pretty bad on the grass, but that was my first EVER fully complete animation, with horizon, skydome, grass, main feature, texture, and background objects. Ya I know I'm pretty new, so i wasnt expecting it to be amazing. But I DID spend maybe 5 hours on it. Personaly it is the greatest thing I have every done. So a great personal feat shall we say it is. thats all.


duller verson

Posted Image
Offline Profile Quote Post Goto Top
 
techwizrd
Member Avatar
Magister ex Machina
[ *  *  * ]
astropirate
Jun 3 2008, 05:49 PM
you wrote the second part your self?
I wrote the whole thing on my own.
My DeviantArt user page Please comment and add me to your favorites!
Posted Image
Offline Profile Quote Post Goto Top
 
slurpee123abc
No Avatar
Slurpee
[ *  * ]
The problem with AI is they can't fluently connect different ideas together. Basically, it means if you're talking to them, they won't put together what you said and mix it with what they already know, then come to a conclusion. I have been talking to some well known AI (A.L.I.C.E), and it appears Alice cannot at all connect previous statements together, and then problem solve. THATS IT, PROBLEM SOLVE. I have not yet encountered an AI with problem solving skills. Thats means inventing the skills. That mean the program rights its own lines of code ITSELF because it is problem solving by ITSELF and not with the help of a programmer. Sorry, I have to go now. I will probably add to this later.
Edited by slurpee123abc, Jul 7 2009, 10:45 PM.
Offline Profile Quote Post Goto Top
 
1 user reading this topic (1 Guest and 0 Anonymous)
« Previous Topic · Programming · Next Topic »
Add Reply