Proponents of the unlimited possibilities of artificial intelligence proceed from the hypothesis that once the functions of the brain are fully understood and correctly understood, they can be encoded and put into a computer.
Many groundbreaking projects in artificial intelligence represent attempts to create a thinking machine. They are based on the idea that the functions of the human brain are limited to encoding and processing multisensory information. In other words, their authors proceed from the hypothesis that once the functions of the brain are fully understood and correctly understood, they can be written as a code and placed in a computer. Microsoft recently announced that it intends to spend a billion dollars on a project with this goal in mind.
However, until now, attempts to create a thinking supercomputer have not even been crowned with initial success. The multibillion-dollar European project, launched in 2013, has actually been recognized as a failure at the moment. In a modified form, it looks more like a similar, albeit less ambitious, American project that develops new software tools for scientists studying brain data, instead of trying to model it.
Some researchers still insist that modeling the thinking processes in neurobiological systems is the path to success. Others consider such efforts doomed to failure because they do not believe that thinking is in principle calculable. Their main argument is that the human brain integrates and compresses multiple sensations, including sight and hearing, which simply cannot be handled the way modern computers do, perceiving, processing and storing data.
Living things accumulate experiences and sensations in their brains, adapting neural connections in the active process of contact between the subject and the environment. In contrast, the computer writes data to short-term and long-term storage of memory. This difference means that the brain handles information differently than a computer.
The mind is actively exploring the environment in search of elements that will help find a way to perform this or that action. Perception is not directly related to data obtained with the help of the senses: a person can identify, say, a table, from different points of view, and he does not need to consciously interpret the data for this, and then query memory whether this template can be created using alternative representations of any object that was previously identified.
Another point of view boils down to the fact that the most mundane, mundane memory tasks involve several different brain segments, some of which are quite large. Skill learning and experience is accompanied by reorganization and physical transformations in brain tissues, such as changes in the structure of neural connections. Such transformations cannot be reproduced in a computer with a fixed architecture.
A recently published scientific paper on this topic has highlighted several additional reasons why human thinking cannot be calculated. A thinking person is aware of what he is thinking. In other words, he is able to stop thinking about one thing and start thinking about another, no matter what stage of thinking he is at. But this is impossible for a computer. More than eighty years ago, British computer scientist Alan Turing came to the conclusion that there is no fundamental possibility of proving that a computer program can stop of its own accord, while this ability is one of the fundamental to human consciousness.
Promotional video:
His argument is based on a logical trap in which there is an internal contradiction: imagine that there is some general process that can determine whether the program it is analyzing will stop. The result of this process will be either "yes, it will stop" or "no, it will not stop." It's pretty simple to understand. But then Turing imagined that a skilled programmer had written code that included a validation process with one key element: instructions to keep the program running if the answer was "yes, it will stop."
Starting the verification process of this new program will inevitably lead to the wrong result: if it determines that the program will stop, then internal instructions will tell it to continue running. On the other hand, if this "stop checker" determines that the program will not stop, the instructions will immediately give the command to stop. This is completely illogical, and Turing concluded that there could be no way to analyze the program and be sure that it can stop itself. Therefore, it is impossible to be sure that any computer is capable of competing with a system that can stop its train of thought and switch to another line of thinking. It is the confidence in this ability that is an integral part of thinking.
Even before the publication of Turing's work, the German quantum physicist Werner Heisenberg showed that there is a clear distinction between the nature of a physical event and the conscious perception of that event by the observer. Austrian physicist Erwin Schrödinger interpreted this argument in such a way that the process of thinking cannot be the result of a physical process, like a computer, which reduces all operations to basic logical judgments.
These ideas are supported by the results of medical research, which indicate that there are no unique structures in the human brain that would be responsible exclusively for thinking. In contrast, functional magnetic resonance imaging shows that different cognitive tasks trigger different parts of the brain to be activated. This led neuroscientist Semir Zeki to the conclusion that "thinking is not something unified, while there are many different thought processes distributed in time and space." Modeling the unlimited capabilities of the brain is a problem that, in principle, cannot be done by a computer, which is a finite system.
Igor Abramov