This study investigates the behavioral differences between machine agents and human participants when they play the same game. We aim to experimentally clarify the advantages and disadvantages of agent-based simulation (ABS) and gaming simulation (GS) and to identify the suitability of each method for various types of research topics. A multi-player game, which models the energy technology selection and price competition of energy companies in a liberalized market, was implemented via ABS and GS under two conditions: with and without carbon tax conditions. The machine agents identified better strategies of energy technology selection in the without-tax condition, whereas the human participants identified better strategies in the with-tax condition. In terms of price competition, the behaviors of machine agents were adaptive and rational, whereas those of human participants were not. These results suggest that the ABS is suitable for determining relatively simple strategies and investigating the adaptive behavior under strategic situations; conversely, the GS is suitable for investigating the real behaviors of humans in relatively complex situations and for inferring the socio-political failures caused by their psychological aspects.