The Trend of Artificial Intelligence Technology - Discussion on the "New Observation Forum" discussion artificial intelligence reply
Author: Leuven Issue Date: 2/21/03 Article ID: 20030025
(I. Introduction
Where is the breakthrough of the next productive leap? From Xinhua News Agency, Yang Jun issued a report of "Artificial Intelligence Technology" reported that more and more scientists are currently hoping in artificial intelligence. They believe that artificial intelligence will bring an unprecedented technology revolution.
This prophecy is mainly from the following two artificial intelligence attempts:
The first attempt was in 1997. The world's number one chess master Casparov and supercomputer "deep blue" have to bow his head, so that the human brain first tasted the taste of failure in front of the computer.
The second attempt is a computer software system that the University of Lens, USA has developed a computer software system that can be written as a simple novel. Researchers say that church computer write novels, chess, etc., computers, etc., computers, etc. They pointed out that computer-playing chess only involves control of simple symbols, while the "narrative" and "organized story" required by computer write novels is closer to the thinking of human brain. With the development of technology, there may be a more talented intelligent machine writer than Shakespeare a day. Rodney Brooks in the Artificial Intelligence Laboratory of the Massachusetts Institute of Mira Institutes, even intended to make his robot COG at least a child's intelligence and ability.
Since these progress seem to have seen the beautiful prospects of artificial intelligence technology. In order to achieve artificial intelligence, people have focused on simulating artificial intelligence and utilizing new technologies using existing computer technology.
According to Yang Jun, the simulation of artificial intelligence with existing computer technology has achieved great results. For example, the development of aided design, translation, retrieval, drawing, writing, play, play, intelligent identification, mechanical operation, etc., has taken a step toward intelligence.
In order to make the computer function more close to the human brain, pay attention to the research of artificial neural networks, which seems to be an important step toward the intelligence of the computer. Some people think that artificial neural networks are new areas of future electronic technology applications, and future intelligent computers may be von of the host. The combination of Nonman Machine (storage computer) and artificial neural network as a smart peripheral.
Furthermore, it is to help the new computer technology to drive manual intelligence. At least three technologies are now likely to trigger a new revolution, they are photon computers, bioacies and quantum computers.
The computational speed of the photon computer speculates that it may be 1000 to 10,000 times higher than the current supercomputer. A quantum computer with 5,000 left-to-left quantum bits can solve the problem of traditional supercomputement to solve the traditional supercomputer within approximately 30 seconds. Relatively speaking, bio-computer research is more reality, US Wisconsin - Madison has developed a more complex DNA computer. It is reported that the amount of information that a DNA can store can be comparable to 1 trillion CD disc. In these speculations, people's development and implementation of artificial intelligence are of course optimistic.
The author is more concerned about the implementation of artificial intelligence technology. Through reading Yang Jun's report, I remembered a "new observation forum" before the "April Fool" is very meaningful, it can be said that netizens benefit Not shallow. Now make some sorting and commentation, for further discussion of netizens.
(2) The essence difference between the map spirit and the current computer
Netizen "April Fool" on January 29th in the "New Observation Forum" posted "Talking Tuli Machine and Machine Thinking" (2000/01 / 29 / 21407.html), responding to netizens "Miss Xiaoshu". From the reply from "Apriloises", "Miss Xiao Jing" has made such a question to "April Fool": A Israeli scientist believes that life is a "Tuling" computer. It is said that the map spirit is more "universal" than the current computer, and any calculated problem can be calculated in theory.
"Miss Xiaoyue" and then asked:
1 "Figure Ling Machine" is different from the essential difference between the computer;
2 Whether "calculated" is determined.
It may be due to the "April Fool" and "Miss Xiao Jing" is a "new observation forum", and the lambed lady is alive and active. The problem of considering the problem, "April Fool" is difficult to answer her questions, but there is a lot of trouble. Jokes (see Ting Ting: "Mathematical Model on Genetic Cryptography", "New Observation Forum", 2000/01 / 28). Now in the face of some questions raised by "Miss Xiao Jing", the spirit of "April Fool" seems to be a bit nervous, speaking is particularly cautious, logic is particularly precise, and the thinking is shallow. "April Fool" said:
In order to answer what is "Figure Ling", you must first figure out the concept of "limited state computer".
Definition 1: There is a state computer - a limited state computer is called "limited" means limited in two ways:
First, it only has a limited number of states;
Second, its time process is "discrete". Although people can argue that the time is continuous flow, the computer is discrete because the computer is running.
State - basic data characteristics of the computer.
Department: All current human digital computers are finite state computers.
Certificate: Proof
Theorem: An unlimited finite state computer will enter the circulation state if there is no external stimulus.
Certificate: Proof.
The narrative of "April Fool" is really logical, through these "definitions", "system", "certification" and "ore "are a clear two children's concept of" limited state computers ". Now "April Fool" began the second definition and reasoning.
Definition 2: Figure Ling Machine - Tu Ling Machine is the theoretical model of Alan Turing in 1936, which is considered an infinite state computer prototype.
The map spirit is assumed to be connected to an infinitely long tape, here, "unlimited long" is considered "unlimited length", or potentially infinite long, not necessarily meaningful.
The trace of the map is equal to the basic unit of "checkered". The map spirit is able to scan the paper once in each unit time, this operation is done and only one of the five things below:
1 Write a symbol to one side;
2 read a square symbol it written on the paper strip;
3 It can remember the symbols on the paper tape it read;
4 It can wipe it out of the symbol it written on the paper tape;
5 It can move forward and backward moving paper with a square length.
(Note: One of the above five operations is considering completed in a discrete time unit)
Because the paper strip is assumed to have an infinite length, the ability of the map spirit is greater than any limited computer. One of the things that the map spirit can do is to simulate any limited state computer. This is because the programs of any of the limited state computers can be compiled into a limited number of code, which can be written on the paper strip, and the map spirit is calculated using them, just as the limited state computer simulated. Perform this program to respond to its excitement of the outside environment. Thus, this "computer" exists in the map spirit in the form of a string "digital", which is thus called "virtual computer." This virtual computer is the most complete simulation of world computers. This complete simulation is called "imitation" emulation. In addition to simulating simple computers, most actual computer simulation other computers are not imited, which is limited by capacity and speed. However, all actual computers can be imitated by the map forever. The map spirit can be imitedly effectively followed by another map. In fact, it can be proved that there is at least one Machi machine to effectively follow all the computers.
Under the guidance of "April Fool", netizens have known the history of the "Tu Ling", especially for the performance of the "Chartering", and the simplicity of the map is simple. With an infinite length), you can simulate any limited state computer. Look at the third definition and reasoning of "April Fool".
Definitions 3: "Universal" Tu Line - a dragon machine, it can emulate all of the map spirit, including itself.
Theorem: There is an unlimited number of computers, which can be completely equivalent to one universal chart, so that any computer can be imitated.
"Apriloons" used three "definitions" answered the "Little Miss" proposed "'Tuling'" and the current computers of the computer ". The answer is very clear, but "April Fool" said that "the" May Not Astrusted Miss "may be fear that the reply of the" mathematical model of the genetic cryptography "is afraid.
(3) How is "calculated"
The second question of "Miss Xiao Jing" "Is' how it can be determined", "April Fool" also made a serious answer:
As for the "Calcible" problem, I want to refer to the "Solvable" problem. In the field of artificial intelligence, it is defined in this way: a problem is called solution, if there is an algorithm to solve it.
It is generally unclear whether a problem is given to the judgment rule. For example, how do you determine if there is a bugs when a program is up to millions of rows or more? You can use limited tests to it, but the scope of these tests is far from covering each conditional path of the program, so it is basically unsubueable if there is a large program without BUGS.
Another example is the so-called suspension problem, this problem is "can't" judgment the solution, because it involves "self-reference", just like the "Converse Paradox" in the collection. Limited to the space, the description is omitted.
American mathematician Baker raised dozens of abstract computers, they are equivalent to Wanneng Charter (see: Becker, Gary S. 1981, "a Treatise on the family". Cambridge, Ma: Harvard Univ.Press In these examples, there are two famous (abstract) machines, a Billiard Ball Computer, a "life game computer", life game computer is the British mathematician John. Kang John Conway, its game rules make us easy to think about the imitation of the map foreground.
Since Tuling puts forward his ideal machine for decades, people have tried to find a (theory) machine, which can do things that the Mandi Machine can't do, but they have failed, there is a conjecture: do not exist Machine. The essence of this conjecture is the so-called "Strong Artificial Intelligence Postulate", that is, human beings can eventually manufacture "smart machines".
This problem is a key reason for the study of artificially intelligent research. The focus of the problem is mainly due to whether the thinking can be restored to a different program (Algorithms), and ultimately the operation of the computer program?
(4) Tuling judgment and Chinese house test
Artificial intelligence research is stagnant, making people have to re-examine some of the judgment principles proposed in Tuling. "Apriloce" describes the Tuling Judgment and China House Test in the reply. He said:
However, what is the smart machine? How to determine if a machine has "wisdom"? Tuling puts forward a famous judgment principle. This principle says: If a person uses any string of questions to ask two objects he can't see, one is normal thinking; one is a machine, if he can't go after a doctor If you find the essence, he can think that the robot has a "wisdom".
Obviously, this judgment is the objective judgment criterion of "thinking" in the study of scientists who study artificial intelligence, in other words, because there is no way to accurately define thinking, or "awareness" object performance Instead, we will come to a shallow judgment. Here, we need to explain:
We are very clear about our thinking. Although not most people can accurately define it, they are also very clear that they mean the meaning of "I", and can also think of others from other people's words and their writings. There is also the self-consciousness of "I", in fact, everyone has already imposed, and it is not convinced of a "axiom": that is, with my similar thinking ability with me. This "experience" can be applied to human beings, but obviously rejected the animal and machinery with human heterogeneous animals.
However, human beings want to create a "thinking machine", so I will come out, you can know what you have made, have self-consciousness? The list of the tops of the above T. (test) is to be aligned.
However, there is still someone (including me, see the "I think I am", one of the sequences, one of the sequences) is not satisfied with the judgment of the Tuling, and it is considered that this does not really determine whether there is a different thinking. Among them, the American philosopher John Searle proposed a famous "Chinese House Test" for this purpose.
"China House Test" said, let us imagine, a person, such as "I", that is, Selsel, is closed in a book, these books can be imagined into a computer code, the machine is considered Has passed the Chinese map spirit test.
We know, perform the operation of any program, which is equivalent to open a book, read its content, copy its content on the notebook (can be seen as a newly printed book), wipe it on the notebook Some words or sentences copied from the book, remember some of its content, jump from a book to another book.
it is good! Now, it is assumed that someone in the door is plug in the door, I, I, a person who doesn't understand Chinese, I will check the book after getting the note, because the book is enough (universal chart forever), so I Can get the answer, and write it with Chinese (also the book, you should say that this is called the Chinese symbol I don't understand), then you will hand out to the person from the door, so I repeated this process, I People who don't understand in Chinese, obviously don't know the meaning of the paper, but let the Chinese people outside the door mistakenly think that I am also a person who knows Chinese. According to the map spirit proposition, "I" will be judged by the Chinese in the door. If I am a machine, I will be understood by him to understand the smart machine with understanding Chinese. However, I don't know. understand!
From this imaginary trial, we see that "understanding" is not the nature of the computer to simulate. It is too small to judge whether the machine is not enough to think about the machine. The essence of thinking is too small. (See: 1. Colin McGinn, 1991, "The Problem of Consciousness", Blackwell, Oxford, Cambridge; 2.howard Gardner, 1985, "The Mind's New Science", Basic Books, New York
At the end of the "April Fool" reply, talked about "Miss Xiao Jing" "There is a Israeli scientist who believes that life form is a" Tuling 'computer ". He said:
If the genetic cryptographic system of the organism is a programming "computer" representing the abstract logic code represented by the DNA's molecular group chain, it is clear that no matter how complicated, it is still a limited state machine. So why will the Israeli biologist think it may be a map forever? I guess: may be because the DNA password system (if it is understood "paper strip" in the sense of figure), it can constantly constitute a new, reinleased code (ie, the symbol on the paper strip) in different ways. In this way, the length of the "paper strip" can be regarded as infinite, so this machine does not turn into a "map spirit"? (See the definition of the foregoing charter)
I boldly guess: If you use this idea to understand "thinking", thinking may be a trace of paper belt with a number of "potential" codes! The definition of the foregoing charter actually implies the picture of the map, the paper tape may be infinite, not an infinity.
If this conjecture can be confirmed or partially confirmed, the Tulex proposition must be modified.
The reply of "April Fool" can be said that I have answered some questions from the artificial intelligence technology proposed by "Miss Xiaoshu". The netizen "confused machine" thinks "The masterpiece of the foolish brother, let us also benefit the lot." "
(5) There are still some other endless
The netizen "paste machine" in "The masterpiece of the foolish brother," 2000/01 / 29 / 21427html) reply to "April Fool" "图 灵 机 就 就 就 就 就 就 就 就 就 就 就 就 就 就 就 就 就Biological genetic cryptographic system is a programming "computer" composed of the abstract logic code represented by the molecular group chain of DNA, he said:
I think someone thinks that the organism is a map forever because the DNA is limited, but the molecules that truly bear the life function are primarily protein. The vast majority of proteins are functional because it has a three-dimensional structure of Unique. Proteins should be implemented in life activities, mainly by protein-protein, protein-other biological macromonomical interactions. It is now found that this interaction is often complex. It seems that these interactions are continuous, not discrete. And this continuous may not be possible, so it is more complicated than the map. Further, "paste machine" proposes the following questions:
April Fool brothers may exist in the mystery of life in an uncommon, and it is considered very deep. If this is the case, the secret of life (especially thinking) is related to the paradox, Goetler theorem, etc., even the same time structure. I think the time will there be a structure?
I want to ask the question to teach the brother is: Infinite in addition to the number of infinity, in the fact that the number is equivalent, there are some other endless, have there any special structure? I know a structureless infinite is to repeat one-third of the discredition in a line, and finally formed a "potential" in solid equivalent, "measure" is a collection of zero. What is the uniqueness of the "probability" distribution defined on this set, can the real world related characteristics?
My ignorance is: The purpose of the topology is to find the condition and simplify these endless structures. If you can't simplify, what is the way to master these endless characteristics? Because if thinking, the secret of life is in an infinite infinity in the brother, what is the endless?
For the "paste machine" "Inferential Infinitely Infrequency, in the fact-equivalent of the number, there are some other endless, there is no special structure?" Netizen "is even for" first "fool "Learn" on "Two Little Questions about Real Variable Functions" (2000/01/29 / 21470.html)
"The uncommon infinite" of the equivalent is also called endless. There is also a greater potential than the continuous situation. For example, the collection of all real functions of [0, 1] closed interval is larger than the continuous system. In general, if a collection is n, the potential of the new set consisting of all subsets is larger than N. This is a theorem.
Regarding "Measure" is a collection of zero, "occasionally" replied:
This is a famous "Cantor), which is generally defined in [0, 1] closed interval (of course, can be promoted). This collection has a lot of wonderful nature, for example, it is a decent set (ie any neighborhood of each point in the collection contains an open range without any point in the collection), and is a complete set (ie Each point in the collection is gathering. The construction method of Tombsake is used to make the above properties (complete, laminated), a set of continuous conditions and the measured measure is greater than zero. It can also construct some strange antique using the Kangli collection.
Finally, "I have a" I have a "confused machine" if you are interested, you can find the textbook of this real-changing function. The paste machine accepts this suggestion, because when the "real letter" is used before, although it is also known that there are many collection trends greater than the "continuous", the corresponding relationship between these collections and the real world can not understand. "Pair Machine" remembers the "mathematical base" class, and the beginning is the continuous hypothesis, and later is Russell, Gendel, etc. But these are also given books.
"Blood Coator" replied in "Answer in" Thank you, "(2000/01/30 / 21607.html) reply," Even "also proposed:
Just, you mentioned the famous Cantor set, which can define the third probability distribution (except for continuous, discrete distribution). I don't know any possible application of this distribution (in other words, what is this distribution? "Even" reposted in "About Singular Distribution" (2000/01/30 / 21698.html) to answer the "Pair of Machine" here:
The so-called "third probability distribution (except continuous, discrete distribution) is a singular distribution. The wider distribution distribution function is a continuous function, but it cannot be represented by an unordered integration, so there is no dense function. Zheng Weikou, the Mathematics Department of Nanjing University, and the fourth chapter of the textbooks prepared by Wang Sheng Wang set out a simple singular distribution function example using the Cantor set special structure.
The theoretical value of singular distribution functions is obvious. As for if there is a specific present, it is in line with the singular distribution, and I don't know.
"Apriloons" in the "Blue Machine" (2000/01 / 30 / 21584.html) appreciated the answer of "Even", and considered the question of the first paragraph questioned by "Pair", has reply It is very good, he is not awkward. He only added a little, it is the "continuous assumption" problem.
Continuous hypothesis ----
There is no collection of numbers, and the genre MIU is satisfied:
A strictly less than MIU strict than C
A: August,
C: Continion,
Miu: Guess the collection of collections.
We know that all the number of sequences (including poor numbers) can also be called "can be" A; all the numbers corresponding to all the real numbers between [0, 1] Related to "Continion" C, the continuous C is existing than the big potential, such as all the number of real functions defined on [0, 1] is a big potential than continuous potential (I even said that it is very clear).
The problem now is: Whether there is a collection of potential MIU, meeting:
A Condor Guess, this collection does not exist, this is the so-called "continuous hypothesis".
Continuous assumptions have been proved to be proven by the active logic from the previous year.
For the question of the second paragraph of "paste machine", "fool" can't answer, but I feel that the problem is very inspiring. The meaning of life, especially the meaning of thinking is a very major topic. At present, human beings have also mastered, but also scientists, philosophers can gradually reveal their essence.
(6), calculated reference should be a philosophical definition
Netizen "One side view" although it is a bad thing for biology, but for the map spirit, calculated, etc., some of the views on "April Fool" "Talking Machine and Machine Thinking" Talking about the charter "(2000/01/29 / 21453.html) did a tip:
Let me talk about what is calculated. The so-called calculated calculated actual should be considered a philosophical definition. It is popular if there is a mechanical process, if we give an input, this process (or machine) can give an answer within a limited step, then this problem is called calculated. Obviously this definition is extremely unstrusted, so some people try to give strict definitions from this century (it should be the last century). One class in the earliest consideration is "recursive function", now known as "original recursive function", although its definition is simple, but the calculated function that people can imagine is the original recursive function. (If you remove Goto, While, for, REPEAT statement in existing programming languages (such as c), then the function we may write is the original recursive function), so people guess the original recursive function is the calculated function. But soon, I found a function of multi-re-reconciled, but not the original recursive function. The mathematician has added a "minimum" operator in the definition of the original recursive function, thereby got a generalized recursive function. At the same time, other people are also doing similar work. Church defines a system called Lambda-calculus (Lisp and Scheme language) from Lambda-calculations), Tu Lee defines his map forever. All of these systems have been widely defined. Church proves that all of these surfaces have failed to definitely definitely the same function class, which means that there is some inner thing behind these definitions. Succession presents the famous Chiqi Project: Directly calculated function class is just a generalized recursive function class or a map spirit can calculate function class.
The point to explain is that the title of the Siqi is not a axiom, nor guess, so you can't prove or have no evidence, you either believe it or not. Because the intuition can be calculated, the class is not a strict mathematical definition. However, because the various computational models given later are actually the graph can calculate the function class, there is no one doubting the title of Church.
From the Ski Project, we cannot make a machine that can calculate the macrower cannot compute function.
Also boldly said: The map of the inner endless paper belt cannot be more powerful than the map of the tape paper. The reason is very simple: the map spirit is required to stop within the end of the end, and all the calculated issues are In the case of natural number, even if it is an infinite map spirit, we have used the most infinite paper strip in the calculation.
Finally, talk about the problem of halting problems. I want to translate a downtime problem more appropriate. This problem is: Whether such an algorithm exists, if we enter a program encoding (such as a C program) as input, this algorithm can give output 0 or 1. If the C program can end within a limited step, then the output is 1, otherwise the output is 0. This problem can prove that it is not a map spirit.
"April Fool" This reply to "one side of the viewer", especially the "recursive function", is considered to be a meaningful complement.
(7) Misunderstanding of artificial intelligence
Netizens "Luankan" for "April Fool" and "Blocirculating Machine" about artificial intelligence, he thinks that there are several issues to clarify. In "Clarifying a few misunderstandings about artificial intelligence" (2000/01/30 / 21527.html) reply:
This is an ideal experiment in the Chinese house, which is not a refutement for the definition of Tulex's intelligence. The experimental designer actually made two assumptions:
1. There is no intelligence in a group of unimpedable components. In the experiment, the dictionary, grammar book, the person (or machine) of the book is considered to be unimpedy, so the designer will think that they are also intelligent. But the human brain is composed of neurons, is there a smart in a single neuron? However, neurons are combined but absolutely intelligent. Now the artificial intelligence industry believes that the unmatched component is combined to generate intelligence. This is the basis for the survival of artificial intelligence research. Of course, the monologists or mysteriousists may deny this. However, their pessimistic prophets made by artificial intelligence are unbroken. Second, can the mechanical examination and a dictionary make a chaotic translation? Research on natural language is early to prove that this is impossible. So this experiment itself is not established.
Tuli machine and brain who have a strong function? Computer technology is still developing at rapid development, and there is currently no clear evidence that the computer cannot exceed the brain in a certain aspect. However, in some aspects such as digital computing, the computer has far beyond the brain. At present, research on brain believes that it is unique to this level of neurons. April Fool's imagination through its own inspiration can work simultaneously in an indisparable machine. This imagination is also a personal belief, but if the emperor is raised, it cannot give any actual basis, and serious researchers will not be interested in it.
The map spirit is of course not universal. At present, the calculated studies will calculate the problem, and the complexity of them is divided into the following categories:
Unsolvable: Do not solve, this type of problem is not available to the map spirit.
Undecidable: It is not decided, but can not be given the lower limit of any calculation step.
NP-COMPLETE: The time is more than any multi-phase time.
P: can be explained in a multi-phase time.
The first three types of issues are considered to be inevitable. Even the P problem, if there is only a high-order multi-phase time to solve, the computing resources it need may not be available in the development of computer hardware technology for hundreds of years.
There are many problems with NP or more complexity in life, and the best way to follow the next one. But intelligence is not required to give the best solution forever. If you find a better way of walking under the complex situation, or find the best solution under the simple situation, this is also intelligent. Such intelligence is sufficient to give only a limited computing resource to give a valid or even optimal solution to the question that may encounter. What does this kind of smart need? It needs to understand and trade on a particular problem at a higher level, such as the judgment of the current situation and the strategy that should be taken, and the identification of the simple but repeated mode, and will die in the following chess Shape and the resumption of the willbeet. Therefore, intelligence is actually from understanding and recognition of specific issues in multiple hierarchies, finding solutions to this issue.
"Apriloce" is also launched "also clarifying the misunderstanding of" strong artificial intelligence hypothesis "(2000/01/30 / 21555.html) The first question of experimental designers actually made two hypothesis:
Your words may be considered. You say "The experimental designer actually made two hypothesis", especially, "I have no smart parts combine," is clearly impressed by "strong manual intelligence hypothesis" The words that makes an unsuspecting to Tuling test. In fact, the "Chinese House Test" is only the crime of "reverse example" proposed by the judgment of Tuling proposition. It does not derive "a set of smart parts combine to come without intelligence". The proposition is some of the people in some religious or philosophical circles.
Obviously, people who propose the Chinese house trial did not use the reasoning to introduce the conclusion, but in your narrative, this assumption is given to others.
For the Luankan mentioned "But the human brain is composed of neurons, is there a smart in a single neuron? But the combination of neurons is absolutely intelligent." Answer: As for the brain, it is only the structure of the neuron collection. It can generate thinking of thinking, and it is not the problem that people who can engage in artificial intelligence can answer questions, accurately, is their belief. Their faith is ok, but don't say people who blame the Tuling test in contrast, believe in the people who are in contrast to their beliefs. This is a logic error.
"April Fool" thinks that LuANKan's reply clearly does not understand the key to this debate.
If you want to make your mind, or have a machine with self-conscious? Instead of being "intelligent" in the general sense of your understanding.
Of course, in terms of high-speed development, the computer can do a lot of people can't do anything. But this is not related to what I want to say, the problem is concentrated in a sharp point: the development of the computer is to create a high-speed, high-capacity "calculation" machine that sets the smart head under the panic, or Can you become an independent thinking, with "I" feelings?
At present, I will have the artificial intelligent trammatings to study, I am suspected that the robots can make real meaning. Frankly, I have been engaged in artificial intelligence in recent years. I doubt that the truthful robot does not build a priori color in front of you, there is a paragraph with beliefs, and in: The logical contradiction of "I" copied two laws. (See the "I think I am").
"Apriloons" talked about the second question of "experimental designers actually made two hypothesis" "April Fool's imagination through his own inspiration, can work at the same time in an indispensable memory Machine. This imagination is also a personal belief, but if the emperor is raised, it can't give any actual basis. Serious researchers will not be interested in it. "Answer:
It seems that you haven't taken care of my text well, I don't say it is just as "tea after the meal", it is already satisfied. Faith can't talk, of course, there will be a serious researcher interested it, I will not take it as a topic.
(8), expand the meaning of "China House Experiment"
Netizen "GTI" has some special reactions to the reply of Luankan. He returned "LU brother clearly expanded the meaning of 'China House Experiment" (2000/01/30 / 21558.html) a post. Unfortunately, the disappearance of "new observation forum" (temporary version) is not possible. Now only watch "LUANKAN", "Danta" (2000/01/30 / 21592.html) to "gti" reply to "gti":
I. Artificial intelligence is not to study how the machine is exactly the same as the human brain mechanism, but to study how the machine can solve the problem that the human brain can solve the problem.
Second, do not use the concept of uncustomary definition, such as "thinking" "spirituality". What is discussed here is there is no intelligent issue, rather than there is a problem with thinking. The people in the Chinese house are the same as the computer, all of them are calculated. To discuss the calculation is not thinking, please give you a clear definition of your thinking.
Third, China House issues trying to use logic to prove that Chinese house is not intelligent (do not understand Chinese) without making a clear definition of intelligence. This obviously depends on the assumption that the components are not intelligent, and there is no intelligence assumption that combines unimplete components. It is acceptable to the dictionary, grammar book, and the person who knows the book is not intelligent (not to understand Chinese). But after a hypothesis (uncomfortable components is not intelligent) is not based. So the conclusion of China House issues is logically not established.
I wrote two refuters (from logically and from logical and from implementation), I must introduce the common sense of artificial intelligence. I just have been fortunate to have passed these basic training. If the fools said, it is just a trainer who has entered the door. I still don't dare to write a deeper personal understanding of this issue with the identity of "thinkers". As for the fool's fascination of "spirituality" "spirit", people who engage in artificial intelligence are also fascinated. For "Spirit" where this complex problem, artificial intelligence is currently selected from the study of more specific problems. Of course, use strict logic to give manual intelligence as a high-level research direction guidance. But if someone tries to introduce supernatural mystery factors to intelligence, a "spirit is spiritual, countless" attitude, that is also self-sufficient.
The response to "April Fool" in "April Fool" is "the center of my original post" (2000/01/30 / 21601.html) is:
My original post talked about "Strong Artificial Intelligence Conjectory", which is the same "wisdom" machine with humans, here, "Wisdom" is not clearly defined in guessing, but guessing this The machine has the same "intelligence" as human beings, which implies the "I" consciousness of human beings.
Since "I" consciousness is a concept of philosophy to present, but everyone can appreciate its significance, my article talks about: Can you make a computer with self-significance, Instead of talking about the "artificial intelligence" problem.
"China House Test" is not to talk about the "intelligent" problem of "intelligence", but to refute the insufficient test of the machine to understand the mutual thinking into human thought. If it understands the unrealweight of the intelligence of the general, it is clear that the essence of this (theory) test is not understood, because everyone knows: Computers can do some work or even people can't do. Therefore, if you use this post to answer my original post, I will not refute you this way.
"Apriloce" thought, "luankan" did not really understand the original post of "April Fool". If "Luankan", what is the "common sense of artificial intelligence introduction" to scare people, "April Fool" will suspect whether "luankan" really understands the common sense of artificial intelligence.
(9), the key to the computer without thinking
At the time of this artificial intelligence, the author posted "LI" network name "Fox's understanding of computer artificial intelligence" (2000/01 / 30 / 21611.html), think:
Japan announced the "fifth generation" computer program in the 1980s, but the true intelligent computer seems to still have not stage, and the machinery is only a highly specialized machine in the factory, so-called mechanical maid or machinery Nanny still remains in sci-fi movies.
Why is it possible to make a thinking artificial intelligence? (Note: Here is a machine that is completely different from people in a cognitive and thinking ability, not today's most advanced industrial machinery or "expert system") this is humanity? In the face of these issues, people can have doubts about whether they can think about the machine. In fact, whether the machinery has a thinking since the birth of the computer (1946) has begun to argue. In 1950, the English mathematician Tiring published a "computing machinery and intelligence" article, and proposed a "Thuring Test" ( Turing test, as a standard for determining whether the machine has a thinking ability. The test assumes that there is one person and a computer, and the people outside the room can be conversantly talking to the rooms in the room through the typewriter or screen. The argument of the Thuringia is that if we don't recognize what the housing is a computer, it has to be acknowledged that the computer has the same thinking than people. That is to say, it "understands" thinking.
In the 1980s, the Philosopher West of the University of Berclai published a "mental, brain and procedures" article, which proposed "Chinese Room Experiment" for "Chinese Room Experiment" Briefly, this experiment is a "foreigner" that does not understand Chinese, and the foreigner can work through the English manual. Through this experimental Sir's conclusion is that the computer will only have a syntax (Syntax), there is no semantics, regardless of how the computer program is complex, and it is impossible to have a character of the human brain (ie the real heart). : Intentionality. Finally, he said, "No matter how the brain is intended, this process must not be equivalent to a computer program. Because the computer program itself is not enough to create intention." In 1989, Oxford University Pengrose published a book of "New Mind" of the Emperor, focusing on whether people's thinking can be restored as a different program, and finally return the operation of computer programs? He gave his negation with an existing mathematical system. Thinking and consciousness cannot be restored to the operation of the program, whether they are more than scientific research? For this question, Peng's consistently lacks a key ring from the entire modern scientific structure, so we can't understand the essence of self-consciousness. At the same time, he also believes that this "lack of rings" is not a mysterious thing that is unknown, and can eventually be explored by science.
Why is the computer's operation of the computer and the sport of the human brain? Some people think that the human brain is more "complicated" than the computer. At present, the main memory capacity of a advanced computer is about 10 8th bytes, which is approximately two, three, three, three, three Magnitude. Therefore, "Parallel Computer Revolution" is proposed.
The fox believes that solving the problem of computer thinking, to increase the approval of the computer main memory capacity, the key to the computer is not thinking is that human beings have not fully understood nature. Specifically, the current computer is used by the Leibniz di, which is the root cause of the computer, because the binary and natural do not match naturally.
The reply of "Fox" makes "fool" excitement, replies, "Foxes see it" (2000/01/30 / 21615.html). "Fox" is coming to this "Thank you for your hopes" (2000/01 / 30 / 21628html), and further talk about the view:
The ill person is relatively late in the artificial intelligence. In 1996, he used to praise and wrote the "Leibnitz binary admittance in Taiji gossip." Providing a possible computer for thinking computers. In order to further think about artificial intelligence, the monks began a series of testimony for Leibnitz. There are many inspirations in the test, thereby understanding Leibnitz and KiCher (A. Kircher, 1602-1680).
In the year of Leibniz, the Western Philosopher wanted to build a "universe language" or called "alphabet of human thoughts", and Leibniz finally pinned in Fuxi gossip, but he used 0 and 1 understanding and numbering of numbers are insufficient.
"Leibnitz binary admission is available in the" Leibnitza-binary "," Libnitz, inventory (1703) in the Netherlands, is talking about three values in 1704. And Leibniz is reluctant to accept, this may be that Lai Bunz is too obsessed with Leibnitz binary originating in the country of China. Such as Goblebignz accepts CAZE persuasion, this World development is different.
The key to the computer can think about the degree of understanding of humanity to nature. Now if people do not describe naturally, they will not be manufactured. So the monks have been thinking about the description of them. Today's artificial intelligence is not a problem with computer experts thinking alone, but the whole society ginseng.
(10), binary and natural do not match
Since the key to "fox" does not think about the computer, it is not only in human beings. It is not fully understood that the current computer is still useful. It is the root source that the computer cannot think. . Netizen "TT" in "嘿 All think now can now achieve no technical difficulties" (2000/01/30 / 21693.html) One of the refuse:
If you use a broad thinking concept, anything is thinking; if you use a narrow thinking concept, the expert system is thinking, anyone's thinking can be used as an expert system. Therefore, there is no technical difficulties in achieving any thinking.
What is general thinking? A system has a response to the environment.
"One side viewer" also said to the "fox" opinion, he said in "2000/01 / 30 / 21670.html):
The smart problem of the computer has nothing to do with several entries. Logically, binary and octal are equivalent. From a computer technology, it is now fully possible to simulate any implementation.
It's prior to now I haven't figured out what is smart. More than a dozen years ago, artificial intelligence is hot, and now it is already a dead. When the storage and speed increase in the speed of 18 months, large databases can achieve problems that people before think of intelligence can be solved.
"TT" is very happy, shouting in the reply: "The argument of the bystander can do it" (2000/01 / 31 / 21686.html).
(11), reflecting intelligence with 'dynamic structure'
Netizens "fearless words" believe that the use of binary is not the fundamental difference between the computer and the human brain. In fact, most neuronal cells have only binary states, agree with the "parallel computer revolution", he speaks in "there is a little I don't agree" (2000/01 / 30 / 21682.html) reply:
I agree with this view. Of course, a single neuron of course is not intelligent, but a large number of units are complex and dynamic connectivity achieve intelligence.
At the same time, "fearless words" believes:
At present, the fundamental difference between machine intelligence and human brain intelligence, I think that the latter reflects intelligence with 'dynamic structure'. From the complexity of the structure, the computer is not good and the cerebellum ratio. 'Dynamic' Structure Structure (Self-learning) is not a computer.
"Turing Test" "confirmed" thinking, I dare not agree.
"Fox" is against "The fearless words" back "There are two points to ask" (2000/01 / 31 / 21765.html), first "most neuronal cells" said " Only a binary status "
The neuron cells we know now is really a binary state? Or do we understand it into a binary state?
Then, "Fox" is "reflecting intelligence" with 'dynamic structure' ":
The "Dynamic Structure" and the fox "Leibniz Binary Attachment in Taiji Gossip is the possibility of" now "now" now we begin to discuss 0 = 1 in the binary of Ruth Temptic binary 0 = ( 1) (- 1) Digital conversion relationship is different .0 ≠ 1 is 0 = ( 1) (- 1), 0 and 1 are two mutually unable to convert, while 0 ( 1) and (-1) can be converted, which reflects 0 = ( 1) (- 1) has a living cycle property of natural substance, and 0 = 1 does not have activity, and it does not meet mathematical logic. So Ruth binary is a system that does not have active, its root cause is no negative or third party to participate in binary. It can be seen that a binary system that does not have active is not a natural material movement, how can it create? What is the essential difference? What is the essential difference? " "Fox" answered the "One Sense" reply in "One Lord" (2000/01/31 / 21768.html) reply, he said:
The fox thinks that the key to the computer is not thinking is that human beings have not fully understood nature. The Leibniz diploma used in the current computer is the root cause of the computer that cannot think about, because the binary of Raise does not match the nature. The "binary and eight-in-one" "binary and octal and octal are equivalent". The key to the "Leibniz binary admittance in Tai Chi Gossip is possible" in the article describes how to describe an activated substance. This problem is what the upstairs is "having a little I don't agree." "At present, the fundamental difference between machine intelligence and human brain intelligence, I think that the latter reflects intelligence with 'dynamic structure'. From the complexity of the structure, the computer is not good and the brain ratio. 'Dynamic' Structure (Self-learning) is even more powerful. "
The human body "dynamic structure" and "dynamic" structure of the "dynamic" structure are really understanding?
(12), binary and three-value state
"The fearless words" for "fox", in "Fox Jun, please" (2000/01/31 / 21895.html), I tried the "Fox" asked "The neurons we now know now." The cell is really a binary state? Or do we understand it into a binary state? ":
Most neuronal cells do only have a binary state. This state is usually defined by the membrane voltage. When there is sufficient ionic channel on the cell membrane being activated (typically there is a specific threshold), the cytostom is converted from the RESTING state (0) (E.G.V = -70 mV) to the activation state (1) (E.G. V = 30 mV). This process is called depolarization. It is a typical mechanism for None-OR-ALL. Mother Nature apparently recognizes that digital transmission is reliable than analog transmission. Digital systems are more robust than the analog system. Therefore, a single neuronal cell can only achieve an extremely simple logic function (equivalent to a threshold function) and there is no function of any storage information.
How is the human brain to achieve intelligence? The current science can only give uncommon guess. My understanding is that memory and reasoning is a complex connection between these large number of simple logic units (I am so-called structure), and this structure is dynamic, there is a self-learning, self-improvement.
Artificial neural network theory draws on this idea. Information or knowledge is reflected in the connection weight between neurons, reasoning or calculation (or function corresponding to this network) is defined by the hierarchical relationship between neurons. (I.E. This is a few layers of network, the relationship between layers and layers, the relationship between the same level, etc.). However, the lack of artificial nerve network is that the self-learning function is insufficient (mostly able to correct the weight), because there is no dynamic structure self-assembly function.
Let us compare the computer. The computer is the logic unit and the storage unit separation. The number of logical units is very small (for example, a single CPU). Stored in a non-structured character form.
And people, for example, when we think of 'Hua', it is not just this character. The image, odor, ... and the feelings given by the flowers will be associated with a certain specified form of neuron groups. The Pattern that these associated units and the activated is the meaning of our brain about 'flower'. The information stored in the brain is not isolated. At the end of the reply, "fearless words" emphasized that he or didn't think 'binary' is the most fundamental reason for computer impossible to have wisdom. "Fox" first believes that two people have certain concepts on artificial intelligence issues in "Fox" (2000/02/03 / 22444.html). "Fox" endorses the "fearless words" mentioned that memory and reasoning is achieved by complex connection between these large number of simple logic units (I am dynamic), and this structure is dynamic, there is self-learning And self-improvement possible. At the same time, "fox" believes:
How to achieve intelligence to the human brain is not comprehensive. In recent years, I think a little thinking with my colleagues. We tend to "the current science can only give uncommon guess", and do a series of thinking and expression for the description and expression of nature, two-value descriptions and Three-value description, the two can correctly express the nature, "dynamic".
Many people do not fully understand our thinking, always think that binary and three-on-generation are different, regardless of binary or three-in-one, they can be interchanged. For example, netizens said in the "IMP": The smart problem of the computer has nothing to do with a few feed. Logically, binary and octal are equivalent. From a computer technology, it is now fully possible to simulate any implementation.
This understanding, we believe that netizens are too dependent on "software simulation", and ignore the most basic nature law, namely the relationship between the body and function. The relationship between body and function is very obvious in anatomical and physiological performance. What is the function of the general law? What is the function of what the body should have?
The diode in the current computer is a binary body that accepts the second will, excluding the will of the three. Nature or thinking process If it is not a binary, it will make the current computer's body gap between the body of the current computer or the thinking process, although the netizen is "software simulation arbitrary import" also changed computer two Value body, eliminating the body of the computer and the nature or the thinking process. Based on this, we emphasized in the "Leibniz binary admittance to Taiji gossip to provide" the possibility of "solve the problem of computer thinking, to increase the memory capacity of computer main memory."
"The" fearless words "" What is the neuronal cell we now know is really a binary state? Or do we understand it into a binary state? "Question," Fox "believes:
It has been clear that only two-value state of most neuronal cells is the current understanding of people, which may change as the experimental instrument changes. As people understand the atomic understanding, people believe that the atom is composed of electron and proton. As the experimental instrument is updated and the experimental level is increased, the atomic structure is made of three values.
(13), small knot
From the above-mentioned netizens debate about artificial intelligence, first, it is important to understand that people have different understandings of artificial intelligence. Some people understand artificially intelligent understanding and expert system, which is misunderstood. Xinhua News Agency, Yang Jun reported that the existing computer technology simulated artificial intelligence has achieved great results, but these achievements are the development of aided design, translation, retrieval, drawing, writing, playing chess ..., etc. in some computers. . Netizen "TT" thinks "If you use a broad thinking concept, then anything is thinking; if you use a narrow thinking concept, then the expert system is thinking, anyone's thinking can be used as an expert system. Therefore, realize any thinking There is no technical difficulty. "One side viewer" thinks that now, the essential problem is still not figured out, what is intelligent. More than a dozen years ago, the artificial intelligence is a while, and now it is already unspeakable. When the storage and speed When the speed increase in 18 months, large databases can realize problems that people have previously believed that they can solve it. "April fool" is correct and explained to artificial intelligence, the key issue can not make thinking Or self-aware machine? The computer is of course in high-speed development. Who doesn't know how many people can't do it? But this is nothing to do with the problems we have to say, the problem is concentrated on a sharp point: computer Developing, it is a high-speed, high-capacity "calculation" machine that created a smart head under the panic, or can become an independent thinking, "I" feelings?
Calcible problem is the focus of this netizen debate. "One side viewer" has made "the so-called calculated calculated active actual consideration is a philosophical definition" in "Talk to the Chartering Machine". This is an entrance to the success of artificial intelligence. Since it is a definition of philosophy, people will inevitably return to "known instructions" and "inappropriate theory" debate.
From the reply, the netizens have a tendency, that is, mankind is still not enough to understand the law of nature. "The fearless words" compare the human brain to achieve intelligence, think that the current science can only give uncomfortable guess. Memory and reasoning is achieved by complex connection between these large number of simple logic units (I am in the so-called structure), and this structure is dynamic, there is a self-learning, self-improvement possible. This concept is generally influenced by the book of Pengros, Oxford, and whether people's thinking can restore algorithms, and ultimately attribute as the operation of computer programs? It is negative with an existing mathematical system. Thinking and consciousness cannot be restored to the operation of the program, whether they are more than scientific research? For this issue, Peng's considered a critical circle in the entire modern scientific architecture, and we could not understand the essence of self-consciousness. At the same time, he also believes that this "lack of rings" is not a mysterious thing that is unknown, and can eventually be explored by science.
"Fox" is obviously influenced by Pengros, which is considered that the modern scientific architecture is lacking a key-critical ring can be explored. For this purpose, it is a non-thinking source that Liebignz is made into a computer. "Fox" supports artificial intelligent researchers' research on human neural networks, while thinking that the focus of the study neural network is not how many neural lines have a human brain, but unveil the mechanism of neuronal transmission, neuronal cells are two-value states, or Three-value state or two or three composite status.
For the calculated computable problems of thinking, Leibniz has explored, and the development of thinking is a calculation. He had fantasy to establish a "universal mathematics", it should study "everything that can be accurately specified in the imagination." The main part of mathematics is "combined theory", that is, the science of operation formula. He said: "I found a way in philosophy, reaching the purpose of Cartesian and others reached by algebra and analyzing in arithmetic and geometry, but for all sciences, Lu Lili and Kili used a combination The method of theory has developed this philosophy, but we failed to go deep into its essence, but they pointed out a road. According to this, all existing combination of combination in the world can decompose the limited simple concept, they are better than It is the alphabet of the above-mentioned synthesis, and the correct way to combine the alphabet can re-obtain all things and their theoretical basis. This finding that if God can let me finish, it will be all the foundations I found, it itself It is very important. "What is the way to find a way in philosophy?" Is China gossip figure! It is clear that Leibnitz has been thoroughly monitored by missionaries, and he has evolved into eight "four levels" (fire, gas, water, soil) of ancient Greece. Element (fire, heat, gas, wet, water, cold, soil, dry). Leibnitz in the "Interpretation of binary arithmesis - binary use simple symbols 0 and 1, its use, and its meaning to China" (1703), said that Chinese people also believe that Fuxi created Chinese characters. Despite the long time, it has changed greatly. He can't help but think about the important things related to numbers and thoughts in the arithmetic experiment. If we are more important than Chinese people to dig the Chinese characters, it may know He takes into account the number when writing. Bai Jin Shen is very inclined to study in depth of this road and is likely to succeed in many ways. But I don't know if Chinese characters are close to a certain feature of "Universal Mathematics" in the Chinese characters, that is, all from the concept of reasoning, through some kind of calculation method, can also come from characters, which will help understand humanity One of the most important ways to think.
Reality tells people that Leibniz has failed to find "alphabet of human thoughts", but his thoughts have stimulated many scientists' wisdom. The triangular state of "Fox" is likely to find "the alphabet of human thoughts" in the retrogrante. In fact, the Netherlands of Amsterdam's Cesar Caze talks about three values with Leibnitz communications in 1704, while Leibniz is indifferent.
The concept of "Fox" believes that the world is calculated, but this calculated must be established to deepen understanding of the basic combination of nature, both two-value and three values, how to accurately describe natural problems. This is the basis for the implementation of manual intelligence.
This article is a comprehensive collection of social, which reflects people's understanding and understanding of artificial intelligence. Today's artificial intelligence has become the participation of the whole society. Based on this in this article, there must be some tattooons. I hope everyone can understand that we can understand that our debate that our arms said is just as a "tea after the meal", it is already satisfied. , The belief is more talk, of course, there will be a serious researcher interested it, we will not take it as a topic. To avoid the "pseudo-scientific" hat, it is known that "pseudo-science" is today is the "historical counter-revolution" and "current anti-revolution".
Finally, it is necessary to have some problems and errors in the process of finishing, I hope that netizens will criticize and correct. (Originally published on February 10, 2000)
