Evolution of artificial intelligence — real and hypothetical social threats

The article considers the problem of artificial intelligence evolution in the context of its negative impact on social, economic and political aspects of society. The stages of evolution of artificial intelligence are described when comparing it with human cognitive abilities, the range of performed tasks and the ability of artificial intelligence to uncontrolled self-learning. General problems of artificial intelligence application in job substitution, manipulation of public opinion and training of artificial intelligence on incorrect historical data set are disclosed. On the example of the popular language neural network ChatGPT the development of artificial intelligence is analyzed, demonstrating the theoretical possibility of evolution to superintelligence and probable risks associated with this phenomenon.