The title of this article was invented by a computer.

And it’s true, but in order. While the development of artificial intelligence is going on, I will not be afraid of this word, one-sided - for specific tasks and in narrow areas. No one brings up a computer as a self-sufficient person with a broad comprehensive outlook . And now ...



Immediately make a reservation, the idea, although practical, but the description is psychedelic, so children under the age of Intel I9 are not recommended to read.

This article is about my new method in fostering artificial intelligence. I am writing “mine” in order to lower the level of the event to the ordinaryness of an ordinary human machine. Yes, for 20 years doing AI at a fairly decent level (I made a considerable fortune on it), I am becoming more and more convinced that a person is a computer. And to create AI, you just need to follow your image and likeness. Probably, just as we were created by a more developed computer. Even organic. Amen!

Have you ever wondered how a person’s self-development occurs? No blah blah, but what are the real physiological processes in our brain? I’m not talking about training - this is not self-development, I mean the process of creating exactly the new, that which was not there and that no one could teach .

Why go far, let's just climb into our brains and copy our physiology! (Well, please do not repeat literally with a scalpel.) What prompts us - people to come up with something new, to solve another problem that has not yet been resolved, or to draw a beautiful picture that does not look like any other? It just tears me away from the desire to give an answer, I can hardly restrain decency for the classic beginning of the article - but any information that for a person, for a computer begins with an introduction: exchange of specification codes.

I am writing only because I cannot help writing, thoughts have been running (through) my head for the second day and third night. And do not say that you are arranged differently. Conduct an experiment - lie down, lie down and remember everything that you thought about during your “rest”! Well? What did you think about, what do you think about every night when you go to bed, what do you think about when you go to or return from work? You can not read the article further, you already understood everything: a person keeps twisting his recent problems in his head. His cache is cluttered even when he is trying to relax.

The semantic wind walks in our head, some images come to the brain, variations of solutions to recent problems, words and terms on situations experienced in recent days. We constantly comprehend past dialogues, mentally speak with someone, prove something ... The most intensely comprehended are emotionally colored questions. Particularly sensitive topics keep us awake. And this is only a controlled part of thinking. Then we have dreams - the hypertrophied scenarios of our daytime films.

Well, to tell you the truth, and all my human-computer discovery - our self-development, our ideas come from the uncontrolled work of the brain (I almost said chaotic - this would be a humiliation for homosapience), caused by recent life experiences.

We can’t even turn it off without special efforts. Mankind has come up with an entire industry for this purpose. Strange as it may seem, sports, entertainment and other forms of active leisure are aimed at the desire to relax. Activity and relaxation in tandem sound like an absurdity, but it is with the help of active actions that you can turn off your thoughts. Rather, change their direction, multitasking saves us from overheating. Our brain is so powerfully imprisoned for self-development that someone needs strong drugs to knock out this function for at least a short time. Alcohol, drugs, and other chemical breakers are all costs of our self-development function. But today it’s not about medicine, but programming.

Well? And what should adopt our less developed computer? I’m talking about a regular PC, and not about our organic brothers. But do not leave him at night a cache of everything that we did during the day? Let the PC once again rotate all the information that the person worked with. He will select various options for solving problems unsolved by users, find the experience of using the Internet, evaluate the relevance of answers to the questions posed ... In general, let it work! If a person cannot fall asleep due to a non-disconnectable thinking process, why not load our brother less?

The first thing that occurred to me was to write a neuron for working with information accumulated from human activities. Actually, in the beginning, everything went smoothly: the classification of material is a trifling matter, the search for solutions is the Internet. It remains only to develop oneself at the expense of the same self-development - working on the same information that remains on the computer after my search for a teaching methodology for working with material. The result was a direct Grail of knowledge, training.

Once, a federal audit came to our military research institute to certify employees, and among other disciplines we had to pass knowledge of the Chinese language (well, for a long time to explain the specifics of that office). Personally, I studied French at the university and naturally approached my boss with a sarcastic question - what should I do so as not to let the team down with a low score? To which the gray-haired colonel answered without a trace of embarrassment: you were taught at a respected university, first of all, not languages ​​and sciences - there you were not a school, but you were taught methods of studying languages ​​and sciences. So work hard!

But the Colonel was right, if we consider not people, of course, but faster intellects in cognition. Maybe the computer will master the Chinese, if it is taught to learn? I’m trying to write a neuron that manipulates with one I understand (what I did) cache for my training.

In general, I went into some kind of delirium. I am developing methods for analyzing the information remaining from human activities. Not by solving a specific task in the field of medicine or video analytics there, but trying to process information garbage, not idiocy? Some kind of confidence in the correctness of the path gives me human physiology, in its image I create AI. And here there is an important difference from modern methods - all of them, as far as I know, are trying to study in very narrow specialized areas. And I went along a comprehensive path where the machine learns everything in a row - based on the materials of current activities. Those. this is a real self-improvement without a direct connection with the instructions of a person. Well maybe.

True, to have material from the activity, the activity itself is needed. While I imitate it, doing computer routine work on creating artificial intelligence programs. I’m joking, kaneshno, I have a lot of social networks, household services, all kinds of banking and government services on my computer.
It turns out that the machine is working precisely on what people are currently working on. Any information has relevance, and the machine in my understanding of the correspondence of human physiology should think precisely about what a person thought today or yesterday. But only this does not mean that she will do what a person needs - she must go her own way. Moreover, we do not even know what she will come up with.

But the path must probably be the same as that of human physiology. Well, I can’t help but think about yesterday’s conflict, which means the machine must work on yesterday’s problem. As I think? Each of my thoughts is accompanied by self-esteem - a kind of comparison with the base of experience that was laid in me before. At the same time, self-criticism is connected (well, with adequate people), i.e. there is a certain selection algorithm FALSE-FALSE. And then in general, it seems, everything is simple, I write the same layers of the neuron that the human self-esteem has:

Regulatory - ensures that the person accepts tasks and chooses decisions.
Protective - ensuring the relative stability of the personality and its independence.
Developing - stimulates the person to development and improvement.
Reflective (or signal) - displays a person’s real attitude to himself, his actions and actions, as well as allowing to evaluate the adequacy of his actions.
Emotional - allows a person to feel satisfied with their own personality, their qualities and characteristics.
Adaptation - helps a person adapt to society and the world around him.
Predictive - regulates human activity at the beginning of the activity.
Corrective - provides control in the process of performing activities.
Retrospective - provides an opportunity for a person to evaluate his behavior and activities at the final stage of its implementation.
Motivating - encourages a person to act to gain approval and positive self-esteem reactions (self-satisfaction, development of self-esteem and pride).
Terminal - makes a person stop (stop activity) if his actions and actions contribute to the appearance of self-criticism and dissatisfaction with himself.

To be honest, so far I have managed with four of them, but more on that later. For now, we continue to build the very principle of the computer’s self-development and gaining consciousness. What, after all this suffering, does a person do with recent information? Indeed, you must admit that your own introspection is far from the final process of solving problems. Yes, I all speak computer language! We just go to other people for advice. We are drawn to friends, to authorities, or even to strangers in a bar, to express our soreness and, perhaps, to get advice. This is not some kind of conscious act; our human-computer instinct pushes us towards it. Moreover, it is more pronounced when relaxing or trying to relax. As soon as a person drinks alcohol, the language itself begins to spit out information.The phrase "and talk?" - became a key in the resting procedures of an organic computer.

Well, well, we discard artificial doping, in any way we discuss our personal problems with loved ones, and production with colleagues at work, sometimes vice versa. And pay attention to how our interlocutors behave? As a rule, this is 90 percent, they become our opponents. It does not matter who is right - who is not, the main tendency is that other people automatically question our conclusions and throw us other decisions. This is almost always an argument! It is worth my friend to say that he liked that girl in white, as I immediately notice that she does not have a completely white dress, but rather beige, and even yellow. I have long understood how to get my children to do something or give preference to something - you just have to say that I do not like it. The easiest way is not to go to a boring film for me - to tell my wife that I really like it.

I am sure that under this article many will throw stones at me. Not because they are evil, since birth, humans are antagonistic. In conversations with the same computer people, we are constantly contradicting each other, questioning the conclusions of others, offer our own versions. Not because we are smarter or even understand this, such is our program given by a more developed computer than we are. Well, not I invented it. But I can use it in the education of our smaller brothers - to put the same mechanism in the "iron" computers. We must teach them to exchange information on the principles of a competitive dispute.

Those. you need to find a network of computers working in the same area or at least close to it. To unite these computers among themselves on the principle of advancing at least opposing ideas, at the most - more advanced ones.

And then the most difficult thing so far - someone or something should become an arbiter. Somehow it is necessary to find sound ideas and cast aside false ones. How does this process occur in humans? And here are some patterns. The decision is made by that person or people whose ideas were supported by other computer people - in almost every country there are presidential and government elections. The enterprises are managed by the most successful managers, both in the past and in the present, who have been able to bring positive results. Etc. Those. every person, except for criticism, makes in his brain a comparative analysis of other people's ideas and chooses the most correct, from his point of view. In the same way, we must get computers to weigh the correctness of the conclusions of all other computers, thereby we get some kind of super correct solution. Well maybe,unless we accidentally elect Hitler. Rather, such results will also be, but any experience is useful.

And now DECISION requires break-in in practice. Oh, here I am afraid that the humanoid experience is not the most perfect. Humanity takes centuries to go in at least a logical direction. Life is accompanied by trial and error, wars and disasters. And there is almost no stability: financial well-being is replaced by crises, friendship is a bloody showdown, happiness is generally noticeable only against the background of misfortunes.

The best thing I can come up with if I follow the same human physiology - sociology, is to speed up the process of checking decisions, after all, the computer’s processor will be faster than a human. Once in a billion. And, perhaps, the very modeling of tasks formed on the basis of the cache will allow human millions of years of self-development to pass in a few days. But how close will it be to human practice? And how much artificial intelligence as a rational creature needs such closeness?

I, as a being from the kind of human computers, still try to solve the first problem - it is probably needed more. I need AI to serve my darling, given the fact that Mother Computer nature made us somehow not very perfect. The second one - let it be, he decides for himself ... No, not like that, it’s better to immediately make sure that he does not begin to solve it - it’s safer for humanity.

So, we again ran into the practice of application. If AI is necessary for people, then it is necessary to try on people, and people quickly do not know how to do anything. And this is the main brake on the self-development of an unreasonable computer; it is unlikely that it will be possible to make us quickly experience its ideas. But at least we have reached a serious intermediate result. Now we can theoretically create a self-developing computer without the use of our special efforts. You don’t need to teach him anything, you just have to write a neural network with the main processing layers corresponding to human physiology and sociology:

• Work with cache, introspection
• Dispute with others, external analysis
• The universal choice of the most correct solution
• Testing Solutions
• Development of a new

But this is not so little, because a person is also not able to accelerate to check the quality of the solution, i.e. we bring the computer closer to our level of self-development. We get a sapiens computer.

So, let's start! The following articles, if the local audience shows interest, will be devoted to the practical implementation of this creature.

PS: Yes, about a strange title. I gave the task to the resulting neural network to analyze this article - that's why I am writing about it at the end - and come up with a heading for it. Well, I certainly wouldn’t give one, even my thoughts didn’t come close to that. Decide for yourself - in the subject? And I wanted to call "Self-development of AI - this is human intelligence." But since I am writing about AI, I choose its version.