On limitations of artificial intelligence

摘要

In this paper, a specific component of human intelligence is specified and described—the “truth reaction mechanism”. For this mechanism a methodology of investigation is developed and applied with the result that human intelligence has to be non-deterministic in a sense which can not be modeled by any formalism. The kernel of the investigation is a paradox deduced from the assumption that human intelligence is deterministic in the sense defined and discussed in the paper.