For many decades, not only science fiction writers, but also serious scientists have been worried about the problem: is it possible to create artificial intelligence? At the dawn of cybernetics, the prospects were bright: it was assumed that at the turn of the 20th - 11th centuries, artificial intelligence technologies would reach such a level that they would make it possible to talk about artificial intelligence. It was assumed that such a time was not far off, but the reality turned out to be more prosaic - despite the constant increase in the functionality of computers, so far it has not been possible to create real artificial intelligence.
Problems of creating artificial intelligence
The numerous difficulties faced by the developers forced scientists to concentrate their efforts on more modest tasks, such as creating expert systems that mimic specialists in narrow subject areas. The problems of creating artificial intelligence are primarily associated with traditional information and computing systems with their formal logic, which cannot even pretend to be a primitive brain. A way out of this situation was opened by fundamentally new computers - neural networks. These are self-learning machines that, to one degree or another, simulate the work of the neural systems of living organisms. However, until recently, no one dared to speak seriously about the creation of artificial intelligence. And so, in 2009, Henry Markman, who is the lead researcher and director of the Blue Brain Project,made a sensational statement, stating that the project has achieved impressive results, and by 2020 the creation of artificial intelligence will be completed. A fully functioning artificial human brain will become reality in the full sense of the word. Successful stages of work allowed the scientist to make such optimistic statements. Markman stated that a hologram of a functioning artificial brain should be presented as early as 2020. However, the authors emphasize that we are not talking about consciousness yet, since this area is still covered with obscurity. Successful stages of work allowed the scientist to make such optimistic statements. Markman stated that a hologram of a functioning artificial brain should be presented as early as 2020. However, the authors emphasize that we are not talking about consciousness yet, since this area is still covered with obscurity. Successfully completed stages of work allowed the scientist to make such optimistic statements. Markman said that a hologram of a functioning artificial brain should be presented as early as 2020. However, the authors emphasize that we are not talking about consciousness yet, since this area is still covered with obscurity.
Developments related to neural networks are, of course, classified, since they represent the strategic interests of developed countries. Therefore, information on these areas is very scanty. It is known that the Blue Brain project was launched in 2005, and its official goal was to decipher the algorithms of the mammalian brain. According to the available information, special attention was paid to new areas of the cerebral cortex - the so-called neocortex. It is these areas that are associated with learning new things and adapting to the environment.
The neocortex is made up of structures called neural columns, each containing 1,000 to 10,000 neurons. Back in 2006, in the process of working on the project, scientists managed to create a computer model of tens of thousands of different neurons, which made it possible to programmatically simulate a column of the rat's neocortex. Each individual neuron is unique, but the team managed to identify the algorithm for the operation of this complex system. This algorithm is common to different brains (creatures of the same species). Thus, a common platform was opened on which the work of the brain is based. At the initial stages of the project to create artificial intelligence, the Blue Gene supercomputer was used, however, to implement a program that simulates a full-fledged human brain, processors with orders of magnitude more performance than in the most modern supercomputers will be needed. However, the authors of the project are confident that by 2018 such systems will already be at their disposal.