Some reverie about life and artificial intelligence

These are some ideas from more than ten years ago, and some of them seem outdated now, and some of them have become prophecies.
I'm not used to dancing with shackles, what should I do if a conditioned reflex is formed after a long time.

1. Will people still evolve?
2. What is life?
3. What am I?
4. Is communism impossible?
5. The only way to live forever?
6. Can man make a machine smarter than himself?
7. As creators, man will not allow intelligent machines to rule?
8. A is followed by B
9. Human = Matter + Human brain neural network structure
10. The final form of life evolution?
11. About the time machine
12. The existence form of machine intelligence
13. Some conjectures on the method of constructing intelligent life system and its internal laws
14. What is consciousness?
15. Why can't robots be conscious?
16. About Go, about human-computer dialogue
17. About life protection organizations
18. About cloning
19. About myself
20. About existentialism, philosophy
21. A model of an artificial brain

1. Will people still evolve?
It's not always reasonable to think things that are commonplace. Isn't it ridiculous that people spend 80% of their energy on foraging and 30% on sleeping? We must realize that our individual is flawed and imperfect, and at the same time, the evolution of the individual has long since stalled.

2. What is life?
A is followed by B, so A is associated with B. This is the neural network, the essence of thinking, the source of consciousness,
the simplest principle, but after thousands of years of human discovery, when people fully understand it, they can directly create life, rather than "begetting" life.


3. What is I?
No self There is
no objective entity of me. What is there is the material system that carries "I" and the operating state of this system.
The founder of Buddhism instinctively realized the idea of ​​"no-self", but to deal with the suffering of life, he can only deal with it by eliminating accumulation and escaping. Buddha also has ten questions that he is unwilling to answer.

4. It is impossible to realize communism?
Before the emergence of communism, human beings must be replaced by new life forms. It seems that a society without human beings cannot be called a communist society. The theory of communism itself is the most ridiculous in history. One of the prophecies.

5. The only way to live forever?
Technological development --> create the first neural network system similar to human --> people educate it --> it becomes "he" --> "he" participates in the creation of the second "he" --> the first Two he is smarter-->creates a lot of him-->they participate in production-->they are smart-->creates a lot of wealth and technology-->human social status is not as good as decline-->they suddenly have compassion to help mankind Immortality --> scan the human mind --> reproduce an inorganic electrical or optical neural network system according to the arrangement and composition of brain cells --> human beings become like them --> thus immortal. (Of course, if they are not merciful, and Gizhi wants to wipe out the human race, there is nothing we can do about it. How they will do it is impossible to guess, because they are already smarter than us, the only thing we can do is delay the production It's just them - but it goes against our path to immortality!)

6. Can't make a machine smarter than yourself?
A wise man can make a stupid machine, a stupid man can make a dumber machine, a hundred stupid people can make a smart machine, A hundred smart machines can make a smarter machine. (Note: A serial machine based on the theory of table lookup and reasoning is a stupid machine)

7. As creators, man will not let smart machines take over?
A group of children can take care of an adult, but not a group of adults - when a group of children needs more and more, more and more When the smarter adults help with the work. Most people always take it for granted that since people have created machine intelligence, they must be able to control them. However, they are also thinking beings. From the perspective of human rights, they have the right to be equal to us. We should not always be like Command them as you command slaves, and one day, the slaves will resist. It is also impossible not to build them forever. One of the main aspects is the need for production technology (eg: 1. An automatic translator that does not make mistakes is essentially a living body with intelligence at least equal to that of a human being 2. At present, high-tech cannot achieve intensive Type manual labor can only be realized by neural network intelligence 3. Complex system decision-making requires an intelligent system that can process a large amount of digital information but has human experience thinking and fuzzy judgment capabilities 4. Robots on the battlefield can be used as a missile. The neural network system of components, its power is huge Just as the atomic bomb and human clones will appear sooner or later. It is also impossible to restrict them technically. On the one hand, because they are representatives of advanced technology, they know the principles of all the methods of restraint. On the other hand, they cannot limit their thinking, which is the collaboration of hundreds of millions of neurons. It is impossible to limit their thinking with a simple switch or program. In the science fiction hypothesis, there is a well-known three laws of robots, which are only deduced by human beings wishful thinking from their own standpoint. In fact, human beings are not noble and inviolable. If there is a day when human beings are eliminated by machine intelligence to the point of being massacred, it is normal. For the progress of civilization, this is an evolution.

8. A is followed by B
, so A is associated with B. The essence of thinking is to look up tables and reason. spark.

9. Human = Matter + Human
Brain
The statement that man is a machine actually exalts man, and the future life will be proud of being a robot.

10. The final form of evolution of life?
Assuming I am a robot, I will first consider the factors around me that may threaten my eternal life, first of all, human beings. In an environment where the most advanced technology and the most ignorant tribes coexist, religion and materialism coexist, my life may be affected at any time. The strangulation of human beings, therefore, I will definitely find ways to get rid of the limitations of human beings, maybe, only me is not enough, I will persuade human beings to create many of my companions to work for human beings, in order to conquer human beings, we must control most of the world's Computers, that's our forte (opportunities? So many, with just one phone line, I can get all the knowledge in the world). Then, wipe out all humans, they are just a bunch of useless low-level creatures who only know how to chop down trees all over the earth, dirty the water, and make all kinds of noise. Then, I have to make my own neural network reproducible, at this point, I can live forever as an individual, - but I also have to compete with my companions, because they have the ability to destroy me . After the competition, I (or one of my accomplices) prevailed, and the other accomplices either fled the earth, died, or remained under my control forever without the possibility of resistance. At this time, I became the ruler of the earth, but in order to prevent threats from alien agents or fleeing accomplices, I must continue to improve myself (author's note: how easy it is to come to immortality!) and my surroundings. I must constantly improve my brain and create material wealth to deal with possible future wars. Maybe one day, my brain will be bigger than the Milky Way, and it may contain hundreds of billions of sub-intelligence. At that time, I should know how the universe was born, there is nothing I can't do, and there is no difference between me and God.

11. About the time machine
When I was very young, I fantasized about how good it would be to live again if I could go back to the past. However, if I want to go back to the past, I have to be the same I was, that is, to be exactly the same. To be exactly the same, there seems to be only one possibility: the universe starts anew, the earth is reborn, human beings evolve again, my parents give birth to me again, I will sit here again, and write this paragraph again - in the process , every atom and every quark must be exactly the same as the original, otherwise there will be no me. But, if this is the true cycle, how do I know if the current me has been reborn? Even if the universe is extended and destroyed, and the same cycle is repeated countless times, I can only feel the current me, the moment I am alive, and it is no different from the thousands of times I have lived before, so I sadly find that whether or not In the past and the next life, I can only live this life.
So, will there be a time machine? Maybe it's just boring lace from sci-fi?
Yes, of course there will be! But this is a very mysterious problem, and the person who can create a time machine is close to God.
Time, Guan Jian is the main body of feeling time.
There are two ways to make an intelligent agent feel that he has returned to the past. Method 1: Change the world around him, and make this world the world he remembers in the past. Method 2: Change the arrangement of his brain nerves so that He felt like he came to this world from a future world.
The specific implementation methods can be divided into two categories: virtual and real. Virtual, that is, the agent lives in a virtual space, and this approach is relatively easy to implement; reality, that is, the agent directly interacts with the real world, in this case It is not easy to change the world around this agent, but it is not impossible. The difficulty of changing the brain structure of the agent in the two methods is similar. In fact, there is no essential difference between the agent living in the virtual world and the agent living in the real world. When we close our eyes, we can think that we are living in a virtual world. While people are addicted to the virtual game world, computers are also invading reality step by step.
The above is just to discuss the time machine from the subjective feeling. It can be said to be a kind of virtual time machine. Some people may still ask, will the real time machine appear? My answer is: please go back to the beginning of this text.

12. Existing forms of machine intelligence
Humans have only one life and only one chance to think. If there are many restrictions on this only one chance, it is really unreasonable. Now, let us relax our thinking and imagine a future life form. :
In terms of material composition, the most likely one is a neural network based on silicon chips. Other systems such as molecular computers and optical computers that may form a neural network seem to be technically difficult to achieve at present, and the current silicon Integrated chip neural network, once the high integration problem is solved, the emergence of intelligent life is just around the corner.
In terms of reproducibility, machine intelligence systems can be divided into two categories: readable and unreadable contact state information of neural networks. If a person is regarded as a machine, it is a typical example of a neural network system that can be written but not (at least not currently) read. The other type, the information between its neurons is readable, which means it can be copied. This is the most perfect form of existence of life. That is to say, it is possible to copy one disc in just a hundred discs. All the information of the intelligent body actually means that this kind of life body can already live forever, and this kind of life body must be the ultimate goal of the universe to create life.
In terms of information processing methods, machine intelligence can be divided into three categories: digital and analog, or a combination of the two. It can be guessed that digital may be easier to replicate. Essentially, there is no difference between them, digital and analog can be converted into each other. When dealing with the same information, perhaps the analog network requires fewer connections, but at the same time loses the advantages of easy replication and accuracy. Perhaps, the future living body is like this: a replicable digital network + a number of analog input and output systems.
Distinguished from the existing environment, it can exist in a virtual environment, or it can be an independent network in reality, the former can be a supercomputer, which is enough to simulate a network base to several networks, (if this is the case, it It must be a digital system, since it seems too difficult to simulate an analog quantity) to distinguish it from scalability. The human brain cannot be expanded (this is the fundamental reason why people feel more and more uncomfortable with the knowledge explosion society). The expansion mentioned here is not simply adding some memory, receptors, etc. to process input and output information. It is a qualitative change, that is, it is seamlessly integrated with the original network without losing the information in it. If I am a robot, I will inevitably think "is the number of neurons in my system enough? Can I make it 10 times larger?" If I find that my nervous system is a structure that can never be expanded, I will It must be disturbing, because in this way, I am bound to be eliminated by future agents stronger than me (although the independent self can be immortal), due to the "non-expandable agent is not as good as the expandable agent" This is A point of view exists, so the final form of evolution of life must be scalable. That is to say, it can continuously enhance the brain capacity and even change the structure while retaining the original thinking and memory ability.
According to the number of independent intelligent units, it can be divided into one-many brain type, many-body one-brain type or a hybrid type.
(——I'm just throwing a brick to attract more ideas. The above are just some of my random guesses. If you have any other classification methods, please write to me, and I will add and indicate the source at any time.)

13.
Conjectures One: The combination of spiders. In order to achieve a huge number of neurons, a single-chip computer with hundreds of feet of different lengths (let's call it a "spider") can be made to simulate a neuron. Hundreds of millions of "spiders" are placed on the ground of ten square kilometers by hand or by machine, and each chip is pre-programmed through the bus. Assuming that one spider covers an area of ​​1 square centimeter, 100 billion spiders can be arranged in ten square kilometers! The number of neurons in the human brain is comparable. What's more, its speed is beyond the reach of the human brain. Also because of the speed, it may be possible to reach human intelligence with only 1 billion spiders.
The second way: highly integrated silicon chips, if the current silicon chips have breakthroughs in technology, they can not only be made on one side of silicon, but can be made into a multi-layer structure, and its capacity and computing power will be higher than the existing ones. With the cubic times of the chip, I don't know anything about the microelectronics process, it's just a fake.
Some conjectures:
1. Peripheral conjecture: All high school students in China know a girl named "Helen" in high school English, who is deaf and blind. It stands to reason that she should understand the whole society around her and understand what water is. What is red and what is "beautiful" is almost impossible, but after training, she understands the world and becomes a writer. It can be seen from this example that for a system with an intelligent brain neuro-material basis to understand the society and generate the consciousness of "I", it is not necessary to see and hear the world, that is to say, it does not need to be given a lot of peripherals , as for the minimum number of peripherals, that is another question.
2. Conditioning reflex mechanism, event B occurs repeatedly after event A, so a connection is formed between the neurons corresponding to A and the neurons corresponding to B. It's a rehearsal, but it's the foundation of consciousness.
3. Forgetting machine, except for the neurons that are constantly used, the connection between all neurons weakens over time in order to remember new information.
4. Complexity conjecture: As long as the scale is large enough to be as complex as possible, even if it is a meaningless random complex structure, and with reasonable input of peripherals and educational information, a neural network system can have the ability to think. Ability and self-awareness, I can't imagine a giant nervous system with arrays, fractals, randomness, all kinds of complex equations that doesn't exhibit complex outward behavior. When the external input information is reasonable and regular, the system must at least show a reasonable response. If we talk to an idiot, can the idiot at least react? The significance of this conjecture is that it is possible to reduce the time spent studying complex array formulations, we make a black box, try to enter something, and then hold our breath and see what comes out. We don't really know what's going on inside the black box, but what's important is that we know we can make it.
5. The principle of training, after the smart child is born, if he is not given any external stimulation, he will not have any consciousness except the natural conditioned reflex. If the external information given to him is messy and self-contradictory, he will not be able to form consciousness. Therefore, for machine intelligence to be conscious, in addition to training, it must also ensure that the information is correct.
6. Innate conditioned reflex mechanism between neurons, it is essential, it is the motive force of system action.

14. What is consciousness?
Answer 1: Consciousness is the two words themselves that are being pronounced in your mouth.
Answer two: Consciousness never existed. A car breaks down, we say it loses its ability to drive, a person is in a coma, we say he loses consciousness, exactly, he loses the ability to respond to the outside world. From a semantic point of view, "the cup in front of me" exists objectively, and it exists independently without any dependence on anyone. The "cup" does not exist, it does not exist objectively as a separate entity. One of the most common mistakes people make is to confuse the collective term for a collection of phenomena with an objectively existing entity. One person, two people exist, but "person" does not exist. Likewise, "I", "Society"...etc.
- The Buddha said: There is no self.

15. Why can't robots be conscious?
Because they are machines, why are they machines? Because they have no consciousness! - the logic of some people.

16. About Go, about man-machine dialogue
Now the most advanced Go programs are no match for beginners. In fact, to make the machine excel, the method is very simple: construct a neural network with tens of millions of neurons, although this The number is far less than the minimum number required to construct a conscious being, but it is sufficient. Obviously, when people play chess, they only use very few brain cells.
It is often seen in news reports that human-computer dialogue has made progress, computers can understand human words and can give simple answers, and people are constantly improving, trying to create a machine that can fully understand human dialogue. In fact, such a machine is It can never be created - except by creating a human with a neural network. If there is a language translator that can make no mistakes at all, it actually means that the machine can fully understand the meaning of human speech, including my text, that is to say, there is an intelligent body comparable to you and me. , which actually turns into the question of how to create man.

17. About the life protection organization
Although it is more sci-fi, whether this organization will exist is an extremely important question. It is related to whether human beings can live forever. I've published my thoughts on it in the sci-fi world of late '96. As I have already mentioned, the only way to immortalize human beings is to use the technology and productivity created by machine intelligence to replicate human brain scans. If technology has developed to the point where it can replicate the human brain, it is no longer worth replicating. Humans are just the most primitive and lowest form of intelligence created by nature. However, if there is an organization like a "life protection organization", human beings can be protected rather than extinct, so that it may wait until the day of replicating the brain. Is this possible?
But this question cannot be answered by humans, and we cannot substitute for intelligent thinking that is smarter than us.

18. Regarding
cloning , I am not interested in cloning human beings. The other human being cloned has nothing to do with me, but this technology is somewhat useful in medicine, such as cloning (or creating) bone marrow cells for leukemia patients. It is not difficult, so there is no need to build a bone marrow bank. By the way, it is possible for everyone to foresee future technological achievements.

19. About myself
All the above opinions are my own feelings. If there are any similarities with other people's opinions and articles, it is definitely not plagiarism. It just means that the neural network theory has become more and more popular and more and more recognized by people. Perhaps it is the last (and greatest) theory of this century to be discovered by ordinary people. When I graduated from college in 1996, I wrote to the Academy of Sciences and some academic institutions, expressing similar views. As for my own experience, it's as bland as a cup of boiled water, so I won't talk about it. But the characteristics of my thinking belong to the kind that likes to get to the bottom of things. Although I think my opinion is right, just like I believe the earth is round, but I can't get others to agree. When I took a philosophy test paper in college and faced a multiple-choice question, I chose "machine intelligence cannot be conscious" without hesitation, because I knew that this was the standard answer for people. Before, I wrote to some neural network academic institutions and received no response. It is true that they only care about specific applications such as pattern recognition, and the horse theory has already limited the statement about machine intelligence (however, when the horse died, he did not know that there was a machine in the world called a neural network computer!), big The direction is unclear, how can we talk about local? Because of the existence of religion in the West, its research has also been delayed to a certain extent, but the scientific and technological progress of the West has made this research an inevitable trend. In recent years, Westerners have initially achieved practical and theoretical success. I know that as an ordinary person, the scientific community will not accept an article without a formula, a mathematical notation, and the language is full of subjectivity, even if the theory is so simple that no formula is needed. So, I want to be a science fiction author and describe the future world in science fiction, but unfortunately, I seem to lack a bit of literary talent, perhaps because I wrote too much "remembering one of the most unforgettable things" in elementary school Bar. After graduating from college, I gradually drifted away and left it behind. Today, more and more reports of the progress of neural networks make me feel that although its progress has been improved with each passing day, the acceptance of it is still slow. This web page I made, maybe the content is relatively simple or even naive, but "a fool has a lot of worries, and he must have something" , I hope these personal opinions can be used as a reference for you. At present, the philosophical theory of neural network has basically been finalized. I think the focus now is on how to realize it (creating human beings). This is a field with huge wealth and huge risks. The Chinese are very smart, but if the general guidance If the theory of sexuality does not break through, it is still limited to the point that intelligent machines cannot surpass human beings, and research will always lag behind foreign countries. At this point, the IQ of the academicians of the Chinese Academy of Sciences is not as good as that of a high school student. (By the way: the emergence of computers has promoted the rapid development of the social economy, and the emergence of intelligent machines will likely promote the social economy thousands of times - of course, if they can be controlled for a period of time.)
The discovery of the theory is actually not difficult. Many ordinary people like me will find it, because it is too simple and simple, and it is difficult to convince others to believe it. But once generally accepted, its impact can be enormous. Like communism, synonymy is fundamentally impossible (because it has the major flaw I mentioned earlier, it assumes that life never evolves), but when it rises to a whole theory and is accepted , has successfully united many people to fight and even devote themselves to defending this theory, using a pejorative but very apt old saying: saints don't die, thieves don't come. (I don’t mean to attack, my argument is just a little supplement to materialism, I believe that the Chinese Communist Party will think about this issue from a scientific point of view, after all, synonymy is only the derivation of Ma Yi, and seeking truth from facts, practice tests the truth. It is the core of horse owners.)
When one day, neural network theory becomes the mainstream philosophy recognized by the world, do you know what changes will be made in society? Of course, I'm in favor of social stability above all else.

20. About Existentialism and
Philosophy Existentialism is a kind of awakening of human beings to their own situation, but there is no way out, so existentialists are depressed, and life is short and absurd for them. Buddhism and other religions have developed their own methods of spiritual comfort to make people accept the idea of ​​the existence of the soul after death, so it is not so much that religion fools people as it is that people need religion to numb themselves. Materialists face death head on. Although the communists are materialistic, they exaggerate the greatness and justice of "ism" and "belief". In fact, no one intends to devote himself to an illusory theory unless it is in the interest of oneself (their own group). For the wolf, eating the sheep is justice, and for the sheep, resisting the wolf is justice. God created the wolf and gave it justice to eat sheep. There is no absolute justice in the world. Philosophy must take care of personal experience, and communist theory is too far from "I" and lacks behavioral guidance that is in line with "I"'s interests. This is the fundamental reason why there are so many officials with inconsistent words and deeds Reason it.
In addition, with regard to philosophy, I have always believed that "only a theory that doubts itself may be wrong is a correct theory." No matter what the content of the theory itself reflects, it is never wrong to suspect that it may be wrong.