Prologue

It has been about four months since I was seriously interested in the problem of Artificial Intelligence. At first, the desire was not so obvious - I just wanted to write a small program with natural language interaction as a workout. But the more I thought about this goal, the more complicated it became and, eventually, changed beyond recognition, became more like Artificial Intelligence rather than what it was at the beginning. Having rummaged in a network, I did not find anything useful on this subject and decided to continue to develop my initial idea.

At the moment, the result of numerous applications of analysis and synthesis is a certain amount of information in the form of concepts, partial architecture models, ideas of implementation and forecasts. So I wanted to share with the public.

At once I want to say that I believe in my idea and develop it not where it will lead, but in such a way that it can be realized, and in the very near future. So that the result could be called Artificial Intelligence (or Mind, I do not share these concepts), while stupidly not giggling and not squinting somewhere to the side with a guilty look.

The first article will be devoted to the basic concept, which I, without thinking for a long time, called the “Concept of Human AI”.
')
I guess how such statements can sound for the average Habrauch, so I am morally ready for almost anything!

Legend

After the 56th, when the first concept of Artificial Intelligence was announced to the world, a lot of water flowed away. Scientists, and after them, ordinary citizens were divided into several directions. And each of them has his own opinion on the question, his own definition, his own criteria. Someone is based on the idea that the AI ​​has already been invented, on the contrary, it is impossible to invent the AI. For reference, one could put all these definitions, but I do not see the point in doing copy-paste from the same Wikipedia. It will give nothing. Most people have their own opinion on this issue and will defend that which corresponds to their views.

Therefore, I will use this right and summarize my opinion, what I mean when I write “Artificial Intelligence”. In general terms, this corresponds to the definition of "strong artificial intelligence", which introduced John Searle. It says that the AI ​​must not be a model of the human mind, but the mind itself, to the extent that the human mind is the mind. Yes, the wording is quite confusing, but I think you got the gist.

List of unused literature

Every self-respecting speaker should have a similar section, so I also think - why am I worse! Only the meaning of this section is slightly opposite for quite objective reasons.

Here I would like to clarify my statement from the prologue about the fact that "I did not find anything worthwhile on the net." First of all, I do not consider any of the main areas of research to be promising for the creation of Artificial Intelligence (see Conventions) in the near future. For other purposes - on health, much is already being used.

Some areas of cybernetics were discussed in my first article, " Painting without an artist ." In it, descriptions of technologies are interspersed with my comments, each of which makes it clear why we will not reach the goal with this technology. In the second article, neurocybernetics became an object for criticism. Already, by tradition, it contains a description of my view on this direction and reflections on the topic of why in order to create AI it is necessary to dig in another direction.

Behind the scenes are still ideas of science fiction and various stereotypes. Consider the most popular ones.

So, I put serial models of robots popular in science fiction films of the 90s in the first place among the misconceptions about AI. This also includes most of the robots of my esteemed Azimov and, for example, C-3PO. I combined them according to the following criterion: they are all created with already wired, programmed for certain actions under certain conditions, intelligence, which is either completely unable to rewrite its program (invariant), or capable, but very slightly. In addition, each of these robots cinematic memory.

What is the delusional nature of this image, which is the goal of cybernetics? If these creatures are described as close, then I do not mind - this is quite possible. But this is not an AI. For a weak-minded robot, you can write some algorithm, but it will be very far from the human mind. But to write a program that gives the robot such properties that it passes the Turing test and in general thinks no worse than a man is impossible. It is the same as to develop a program for teaching a person of preschool age, so that in a week (month, year) he masters the knowledge and experience of an adult.

Well, in the second place I have fantastic mega-computers: Skynet, UAV from Stealth, VIKI from Me, Robot, Set from Universal Soldier 2. I think you can continue the list yourself. In principle, they are not particularly different from the robots discussed above, but they are perceived not as something human-like, but as a program that can be run on a computer. Many hope that such things will hang on servers with web search engines, on their own machine as part of any IDE or even as an OS. At the same time, some purely human properties are expected of them: thinking, the ability to solve nontrivial tasks, make difficult choices, and the like.

What is wrong here? Indeed, it looks beautiful and convincing. On the other hand, in such hypothetical systems they want to use only some of the human properties, while others, because of their uselessness, are not welcome and are ignored. For example, I perceive this situation as follows: “I like apples, so bring me a sort of apple trees that would consist of only apples.” In other words, we want to take a part of the intellect, and that this part no matter what is a full-fledged intellect.

I propose to dwell on this. But if such a need arises, some well-known theory or hypothesis will emerge that I did not mention, then it may well be here, among others, in the list of unused literature.

Main provisions

In some way they follow from the previous section. Our Human AI must have the maximum number of properties of its creator. After launching, he should have the same knowledge, that is, almost none, and the same skills that can generally be called the ability to learn. Further it will work autonomously.

The learning process will take place in the same sequence as that of a person, starting with an acquaintance with the outside world, learning to speak, then comes the development and complication of the picture of the world. All these stages should occur in the same way as in humans. Therefore, an important criterion is to communicate with people. Without it, the AI ​​simply cannot develop. It follows that after the AI ​​is created and launched, it will take several years to train it. Strictly speaking, he will continue to continue to study in the same way if the necessary conditions for this are achieved.

One would think that it would be much more efficient to first organize some block of information in a special way, and then just copy it into AI. But the human mind works with information much more complicated than a search engine. The AI ​​must, in the same ways as a person, obtain this information and then streamline it, again just like a person.

I can not yet say exactly how fast the training will take place. After all, the most important information, especially “at an early age,” the AI ​​will receive from communication with people like any child. So at the first stage, this will limit the speed of learning. Further, when he reaches the mental development of at least 15-20 years old person, he will be able to gather a lot of information from books, articles and other text carriers of information. With such a development, this information can already affect his views, give knowledge that he can take and use.

Here is the question - how fast will the training take place? Can the limitation in the speed of human learning be a consequence of the peculiarities of the structure of his mind? How much can you speed up the process with the same functionality and efficiency? The partial models created by me are not yet perfect enough to give answers to these questions. But now we can say that the option, in which the AI ​​will launch on the Internet after the launch, will become wiser within a couple of minutes, and in a couple more minutes - it will launch nuclear missiles, disappears. How we teach him that he will become.

As for training, the question may arise - can the AI ​​outperform a person? Let us forget for a moment the prophecies of science fiction and sensibly assess the situation. In this case, the intellect must be perceived as something concrete. But the specific intelligence of different people are different. There are geniuses, there are idiots and there are other normal and not very people. So, Artificial Intelligence will be able to be on equal footing among not the most stupid representatives of humanity, as well as transcend them. After all, he goes the same way from birth, possesses the same stimuli and abilities to cognition, his memories are also organized. So ask yourself the question, what prevents you from becoming the smartest in the world?

If we create an AI that is identical in structure to the human mind, then we will automatically endow it with all the properties and functions characteristic of humans. This includes feelings, and intuition, the ability to lie and forgive - everything that relates to the peculiarities of the mind.

It is much easier to create something that functions exactly as our intellect, than to pretend to be God and from scratch to invent a system that could develop autonomously (without interfering with the architecture) from the moment of launch. That's when we make the Human AI, when we understand what works, then we will know what needs to be optimized and how, if anything at all, is optimizable.

How will it look like

Everything described in the previous section is nothing more than a philosophy, an idea that should permeate all stages of development and be sure to be present in the final result. What this end result will look like is dedicated to this section.

Platform

We have already decided that we will have the maximum functional similarity in the architectures of AI and human minds. This means that absolutely everything relating to the human mind must also be present in Artificial Intelligence. I emphasize that the similarity must be functional.

If we need something “absolutely” humane, then some problems may arise. A man lives not only with his intellect, thoughts, dreams. His head wears a body that constantly makes itself felt and interferes with our sacred thoughts and dreams. However, without this intervention, we would not be ourselves, as many processes in the body somehow affect our Self. Something affects only superficially, for example, when a bad tooth prevents us from concentrating, but there are also such things whose influence is somewhat deeper like hormones. Let's not forget about innate fears, instincts, etc. If you try to copy all this diversity, you will have to create an exact copy of the whole human body besides the intellect itself, and in the end we will get a person. I personally do not see the point (except self-assertion) in this, and I do not see an opportunity to achieve this in an “unnatural” way. Yet we are not gods and not even a UFO.

Consequently, the level of abstraction of a model (system) must be higher than the level of cells (neurons). In this case, the created model will be platform independent. The priority platform is naturally the computer. The level of abstraction must be such that the whole system can be implemented in an object-oriented language. And since any information can be encoded using text, then, in order to organize the memory, it is only necessary to develop a method of modulation (coding) and markup. The main advantages of this solution are: saving on expensive extraterrestrial technology through the use of conventional computers with conventional hard drives for storing memory and saving on programmers working with an extraterrestrial programming language in favor of the good old C and the like.

Interaction

With the platform decided. We now proceed to the description of the properties that it should have in the first place. I will cite the thoughts of Minski about Artificial Intelligence:

Research in the field of artificial intelligence is aimed at developing programs that solve such problems that a person is now better at, because they require the involvement of such functions of the human brain as the ability to learn based on perception, special organization of memory and the ability to draw conclusions based on judgments.

Here it is - the ability to learn based on perception. In order for the intellect to receive information, something must supply it with it. In humans, this function is performed by the senses. But to reproduce each of them will be difficult. However, this complexity does not come to any comparison with the problem of storing the information received. After all, it is necessary not only to record and reproduce it, but to perform more complex actions on it, for which it should be interpreted and stored in a special way, as it is stored in humans.

Which of the developments in this area now exist? Probably the most important of them and, probably, therefore the most developed is the sight, namely, pattern recognition. But the level of its development does not allow its full use, i.e., as we do. So the first thing we have to give up is vision. For most people, vision is the main source of information, so it needs to find a worthy replacement.

Let us think about why we can now try to organize a normal “human” memory. Perhaps many will say "text" - and they will be right. Although a person does not have such a separate body, we decided that the similarity would not be stupid one to one. Already there are systems for storing knowledge in the form of text. Of course, in this form, as they are now, they cannot be used, since, again, the person uses their analogs differently, but this is already something.

By textual form of communication can be attributed speech and in part - the ear. After a small study of the literature on the education of blind children, it turned out that a prerequisite for the formation of speech is the presence of non-textual information coming from the senses. So far, we have not decided on any of them, but it turns out that one cannot learn speech without information coming from the senses, since words cannot cling one to another and stick only to each other (... and then I remembered the knowledge systems ...). Be sure to do some information in addition to words, so that they have something to connect in memory.

Perception

As noted earlier, it is not yet possible to realize any of the human sense organs, except perhaps only externally. So, we need to come up with a completely new body that will give information about the outside world.

I admit, at first I had an idea to use a virtual world created on a computer in the form of a three-dimensional interactive model. A kind of matrix for AI. Especially since there are a lot of such now, and only one Half Life 2 is worth it. It will be easier than ours, and it will be easier to organize vision. But even here there will be a lot of problems, since we do not know how a person has visual memory, and it will take a very long time to figure it out by trial and error on such a complex model, even for such a relatively simple world.

It is not necessary that the final version of AI be equipped with only one sense organ. At the first stage, the main task of this body is to assist in the development and adjustment of the mechanisms of learning, memory and others, so that they correspond to human ones. Hence the main requirements for it.

It should be simple, but giving ample opportunities in interaction with AI. It should not distract from its complexity from work on memory tuning, forcing it to engage instead in cumbersome recognition algorithms and interpretation methods for subsequent storage. At the same time, the technologies on the basis of which it will be built must be sufficiently studied and developed so that it will not have to be invented to use it.

And one sunny day later, a mile and a half and a half album, I realized that it should be. Vector graphics.

Yes, yes, yes, it is simple, free and has a lot of advantages. Although it is a purely computer thing, it can serve as a simplified analogue of human vision and in part even touch. It gives tremendous freedom of action, since it is possible to form a whole environment from different objects with different colors. Thinking a little more, you can create dynamic objects, allowing them to move, change, interact, give birth to new objects.

Attached to the real world, objects will serve as analogs of toys for a small AI, manipulating with which it will receive various information. These objects can imitate real objects and some of their properties, so that AI can learn their names. In short, the scope for fantasy is huge. With such a system alone, one can achieve much more than the development of a one-year-old child. Indeed, in the vector, you can organize numbers, maybe even music ...

Well, now the portrait can be considered complete. Outwardly, our human Artificial Intelligence will look like a window in which there is an area for drawing, a text field, buttons ...

It may outwardly resemble a hybrid of Corel Draw and Tamaguchi, but in fact it is a great platform for testing and refining various ideas, which, I must say, are already mass.

Epilogue

Those who read to the end and have already forgotten about what was at the beginning is dedicated.

This is not a sci-fi idea and not a forecast. This concept was conceived and developed in order to be able to implement it. I do not deny that AI can be implemented in other ways. It’s just that in my opinion, which I set out in detail as I could, other directions will require much greater expenses both in time and in money.

In principle, the main part of what is written in this article was formed several months ago. So most of the time I did what I was trying to understand how Artificial Intelligence should work (see Legend), how its systems should be arranged, what methods and sciences will help to figure this out. Despite the fact that it is still far from ready-made models, there are still a lot of considerations and ideas. If readers have such a desire, I will write an article on this topic.

Table of contents:

History of Artificial Intelligence, part 1. Painting without an artist.

The history of Artificial Intelligence, part 2. Neural network AI - inevitable or impossible?

Making Artificial Intelligence