"Human, too human." Will we not become hostages of our mind, realizing a universal AI?

    Many scientists and engineers have unreasonably high expectations for artificial intelligence. For some reason, they believe that thinking machines, once created, will immediately solve many of our problems and lead to technological singularity. However, this way of thinking is a delusion.
    Note. I deliberately do not use the word "intelligence", because under it usually means the ability to analyze problems and make decisions based on templates, without understanding the process. In this value, intelligence is inherent in both the Maple program, which solves complex equations, and bees, capable of learning and complex behavior, and many other animals. In return, I will use the word “mind,” referring to human-level thinking machines.

    Do we even need an artificial intelligence? Is needed. Alas, “naked” intellect solves a too narrow class of problems, and sometimes it is very inefficient. Moreover, “naked” intellect is not able to formulate the problem by itself! For example, let's take a game of chess - the machine solves the problem not creatively., spending a lot of resources and using various algorithms embedded in it by engineers. A person solves a problem creatively - he generates new solutions, and using much smaller resources!


    Let's return to our problem. Most likely the first artificial intelligence will be created on the model of the human. For we still do not know how consciousness works, and so far the only available way for us to construct an artificial mind is copying the human. In other words, the thinking machines created by us will model the human brain. And this very important circumstance imposes serious restrictions on our expectations.

    The fact is that we plan to set creative tasks for artificial intelligence . We expect from him breakthroughs in medicine and biology, physics and cosmology, sociology and psychology. Moreover, we believe that he will be able to write books, compose music, draw pictures. In other words, we believe that he will be able to create new knowledge(in a broad sense - including art and science). But most adult healthy normal people do not do anything like this! They are really normal, for the majority determines the norm.

    Being created on the basis of averaged human, artificial intelligence will not differ from it, except for potentially unlimited resources. And it is quite possible that an artificial mind, possessing consciousness, simply does not want to do all this beforehand with the creativity entrusted to it. After all, modern man has unlimited resources in comparison with his ancestor from the Stone Age. He can get any information accumulated by mankind in a few seconds. But at the same time, few people need it, except for scientists and engineers. Most of the information industry is working on entertainment. So it was and always will be, regardless of the amount of available resources (provided that human nature remains the same).

    Thus, creativity is an anomaly rather than the norm. Of course, the consumption of information is in itself a creative process that requires some kind of creativity, as, indeed, life in the modern world. However, such creativity at the household level is necessary for the purpose of survival and is not very similar to studying the Universe or writing music.

    For example, the emergence of musical tutorials made life easier for beginning musicians, but did not increase their number. If a person is not interested in music, he will not deal with it, no matter how easy the process is. Similarly, whatever computing resources the thinking machine has, if it is not interested in the problem of human aging, it will not solve it.

    However, several ideas were expressed on how to circumvent such difficulties. For example, you can program the artificial mind so that it solves the problems we need. However, until we find out how our mind and any mind work in general, until we understand what changes need to be made to the brain at a low level in order to generate interest in music, we will not be able to influence our brain and, accordingly, on its computer model.

    Well, well, opponents would say, we still can not program the mind directly, but we can raise the personality we need. This is certainly a more realistic idea, but, as before, not implemented in the framework of today's knowledge. (And education itself is a form of high-level programming!)


    There is an equally important problem that is usually not talked about - but in vain. The problem is human nature. Man is an animal, and for an animal it is normal to receive physical pleasures and to multiply. And anyone who spends his life creating new knowledge, instead of making a lot of money with the least effort and getting more physical pleasure, can be considered “insane”. In psychiatry, it is not considered normal when one idea captures the mind of a person over the years. And many talented people are really insane or have other mental problems. Of course, such people enjoy their creative process, the problem is that it can be dangerous for their mind or life. As well as the creative people themselves are sometimes dangerous to others.

    Our entire civilization is built by "crazy" people! Everything around us was created by scientists and engineers, artists and musicians, writers and other creative madmen who have dedicated themselves to changing the world, sometimes to the detriment of their lives! And yet, “mad geniuses” are more or less used to it, and it is easier to control a vulnerable person than a car, and you would entrust your lives to a powerful artificial mind if you doubt a little about its responsibility?

    It is quite possible that in the near future we will learn how to design really creative thinking machines, but will they not be as insane as their human prototypes? And aren't mental problems an integral part of any creative process? Questions to which we still do not know the answer.


    In other words, the artificial mind will have the same problems as the human mind. How to avoid them? We need a universal theory of the mind , explaining the principles of the functioning of any - or, at least, the human - mind. Only knowing the principles of the mind, we can design smart machines, not by blindly copying, not by trial and error, but by targeted design. We take the human mind as a basis and improve it or even redesign it altogether. With this theory, we can improve and your mind.

    We need a breakthrough in philosophy, we need a theory explaining how the brain creates explanations.
    David Deutsch, British physicist and philosopher

    Compare the first clumsy aircraft and the current airliners or military F-35! The current aerospace engineering is an example of our good understanding of the laws of physics. Engineers do not copy the wing of the bird, they calculate the necessary parameters, applying the laws of physics.

    Alas, in the development of artificial intelligence everything is a little different - we still do not have a universal theory as, indeed, attempts to build it, which means there is still a lot of trial and error and time ahead.

    Also popular now: