The technological evolution of artificial intelligence in game playing, particularly in chess, exhibits distinct generational shifts. This paper systematically categorizes existing research into three technical paradigms: the Rule-driven Paradigm, the Data-driven Paradigm, and the General Algorithm Paradigm, analyzing their intrinsic evolutionary logicincluding optimization of search mechanisms, migration of knowledge representations, and paradigm shifts in training objectives. Through comparative analysis, it reveals a synergistic evolution between enhanced algorithmic autonomy and optimized computational efficiency. Addressing current research bottlenecks, the study explores potential pathways toward general game intelligence, such as incorporating meta-learning architectures and learning universal game representations. Findings demonstrate that chess AI's progression not only expands the boundaries of game intelligence but also offers methodological insights for designing general decision-making agents. This advancement fundamentally hinges on progressively reducing reliance on human priors while achieving adaptability to complex gaming environments through algorithmic innovationsparticularly diverse strategy generation. This trajectory underscores the continuous pursuit of autonomy and efficiency in AI evolution.
Fang et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: