Chess has long served as a model domain for studying search, expertise, decision‑making, and artificial intelligence. The rise of large language models (LLMs) has revived chess as a controlled environment for investigating strategic reasoning and comparing human and artificial decision‑making. This paper conducts a systematic mapping of recent work covering human players, classical chess engines, neural and reinforcement‑learning systems, LLMs, and hybrid approaches. The final map contains 84 core study families, classified by agent type, strategic‑reasoning stage, and evaluation dimension. The map shows that most literature concentrates on situation assessment, evaluation, and action selection, while explicit planning, explanation, metacognition, and human‑AI collaboration are less explored. LLM research emphasizes state representation and generalization across positions; hybrid systems that combine language models with engines, expert knowledge, or other external structures more frequently provide grounded explanations. Two unresolved distinctions emerge: hybrid systems differ in where and when heterogeneous capabilities are combined, and evaluations that improve human performance do not automatically demonstrate human‑AI synergy. We propose extending the mapping framework to capture these aspects. We argue that chess bridges cognitive and computational perspectives on strategic reasoning and identify explicit planning, faithful grounded explanation, metacognitive calibration, and human‑AI complementarity as key future directions.
Review: The paper offers a comprehensive overview that clarifies existing work and gaps, guiding future research on strategic reasoning within the chess domain.