Artificial Thought Model for realizing flexible information processing

The authors are intending to develop human-like computers which can work in the unexpected situations encountered in the real world and in fluctuating environments. Such computers basically do not have programs as those used in conventional computers. The authors think that an organic coexistence of fixed and speculative processes will be useful for realizing flexible information processing, and the authors have proposed the Artificial Thought Model. The model takes a foresighted approach, which is constructed from forecasting and speculative processes based on forecasted data. The forecasting process is updated by a self-organization system based on, for example, genetic algorithm (GA), neural network (NN) system, or artificial life (AL), in which the real input data of the fixed conversation process is appropriated as target data for the self-learning system. By using the game of tic-tac-toe, the authors evaluated the advantages of this process. The Artificial Thought Model is also constructed from a hierarchical structure of sequential logic. It will be useful if such computers could adapt or extend themselves on their own to cope with fluctuating environment or unexpected situations encountered in the real world.