Current AI agents in wargaming often rely on preset rules and historical data, limiting their adaptability in complex and dynamic scenarios. This highlights the need to better understand human decision-making. This study employs eye-tracking technology to investigate human decision processes and cognitive load in wargames. By analyzing eye movement data, we aim to understand the decision patterns of human players. Six key eye movement metrics are identified as effective indicators of decision complexity in wargaming tasks. Then, we develop a deep neural network based on these metrics, achieving an accuracy of 88.49% in detecting players' current decision tasks, significantly outperforming traditional models like Random Forest (86.0%) and GBDT (83.97%). This research demonstrates the intrinsic connection between eye movement features and decision-making tasks, offering valuable insights for designing human-aligned AI agents and enhancing perception in human-AI collaborative gameplay.

In the field of management research, complex decision-making scenarios are frequently encountered. Intelligent decision-making games serve as an essential tool for simulating such scenarios, enabling decision-makers to evaluate strategies and allocate resources more effectively. However, traditional intelligent decision-making games relying on deep reinforcement learning (DRL) often suffer from prolonged training times, convergence difficulties, and challenges in multiagent coordination. To address these limitations, this study proposes an enhanced game framework that integrates role-based collaboration (RBC) with the environment, class, agent, role, group, object (E-CARGO model). In this framework, agents are first assigned different roles, after which reinforcement learning (RL) techniques are applied for policy training. Simulation experiments conducted on the winning-first platform demonstrate that the proposed method achieves superior convergence performance and agent intelligence compared with conventional RL approaches, effectively mitigating the identified challenges and enhancing overall decision-making efficiency.

Role-based collaboration (RBC) is a role-centered computational paradigm for solving collaborative problems, where group multirole assignment (GMRA) is an important component. This article focuses on leader–member teams, a common organizational structure in project management, and extends the GMRA framework to address two critical challenges. First, evaluating the qualifications of leaders and members is nontrivial due to their distinct responsibilities. To address this, we propose a capability–requirement matching evaluation (CRME) method that applies differentiated mechanisms to assess leaders and members. Second, existing studies mainly maximize overall performance while neglecting task performance balance, which is vital for synchronized progress. To overcome this limitation, we develop a group multirole assignment with balanced task performance (GMRABP) model that incorporates a penalty-augmented objective to maximize team performance while reducing disparities across tasks. Furthermore, two linearized variants, GMRABP-A and GMRABP-B, are introduced to enhance computational efficiency. Extensive experiments and comparative analyses validate the effectiveness of the proposed methods, offering practical strategies for managing projects where both performance maximization and progress coordination are essential.

This paper introduces a novel paradigm for building trustworthy autonomous systems through explanation-driven adaptation. We propose a co-adaptive architecture where black-box neural policies, interpretable policy distillation models and explanation engines evolve synergistically through bidirectional feedback loops. The framework fundamentally transforms explainability from passive post-analysis to active system guidance, enabling continuous policy optimization while maintaining human-aligned transparency. By embedding explanation consistency as a core adaptation objective, our approach establishes dynamic equilibrium between environmental responsiveness and operational verifiability. This work advances the design of safety-critical autonomous systems through formalized principles for explainability-guided adaptation, creating new pathways for resilient human-machine collaboration in complex service environments.

Traditionally, economic models have often been based on fixed assumptions and analyzed within static scenarios. Economists have long sought a more flexible and dynamically adaptive model to find more precise models to represent socioeconomic laws. In this study, we built an economic engine grounded in multi-agent intelligent game simulations and employed reinforcement learning (RL) techniques to simulate economic activities. By establishing diverse agent behaviors across three industries and iterating taxation adjustment policies, we simulated and deduced socio-economic activities. Our simulation experiments confirmed that multi-agent intelligent game simulations, underpinned by reinforcement learning, can to some extent capture valuable economic operational patterns. The AI tax table, derived from the evolutionary economic engine, proves to be more realistic; it can boost overall societal production while ensuring maximal social fairness, thus finding a valuable local Nash equilibrium.

The human–machine collaboration task-allocation problem involves three major challenges: role diversity, capability heterogeneity, and task dynamics. Most existing studies treat humans and machines as parallel units through a static “Human + Machine” additive paradigm, which neglects the evolution of capability during collaboration. Some works “deeply couple” relatively low-autonomy machines with humans, thereby limiting the system’s flexibility in resource scheduling and dynamic reconfiguration. This study analyzes the problem from a multiagent system perspective and classifies execution units into three types: human agents, machine agents, and human–machine collaborative agents, and distinguishes between their independent and collaborative capabilities. Next, we propose the dynamic short-board balance synergy assessment method, which integrates the “short-board” concept to quantify collaboration performance and leverages agents that have low independent but high collaborative capabilities. By incorporating multiple constraints, we establish the role-based human–machine collaboration (RBHMC) model, prove its NP-hardness, and design a multi-level solving approach to handle small-scale and medium-to-large-scale data separately. The experimental results indicate that, compared with “Human + Machine” and “Deep Coupling” models, RBHMC outperforms in task completion rate, resource utilization, and system robustness. An industrial case study further validates its applicability and superiority in real-world settings. Finally, RBHMC’s transferability is validated through vertical technology adaptation and horizontal scenario migration, providing a scalable solution for multidomain human–machine collaboration in complex scenarios.

With the widespread application of AI in workplace scenarios, integrating AI into workflows has become a significant trend. However, most existing studies treat AI as independent agents operating in parallel with humans, assigning tasks in isolation, and failing to fully exploit AI’s impact on human capabilities. This article goes beyond the simplistic division of labor and proposes an AI-augmented collaborative task allocation method, emphasizing AI’s role in supporting human performance. By systematically modeling factors, including individual differences, interpersonal conflicts, technical constraints, and AI’s dynamic impact on human capabilities, we establish a multidimensional AI-augmented capability model to quantify capability impacts. Fuzzy interval numbers and cloud models are employed to address measurement instability and the heterogeneity of individual capabilities. Real-world case studies and numerical experiments validate the method’s effectiveness in scenarios that reflect realistic office characteristics and scales. Furthermore, experimental analyses identify transition patterns in AI-augmented environments, and verify the method’s adaptability to different AI development stages and diverse business contexts. The results provide a new theoretical perspective for understanding organizational resource reallocation driven by emerging technologies.

Wargame has emerged as a preferred instrument for simulating combat decision-making. This paper employs machine learning methodologies to predict the outcome of wargame matches. Initially, we conducted data preprocessing on 335 wargame match replays, extracting and generating features from both macro and micro perspectives, thereby capturing player strategies and operational nuances. This meticulous process culminated in the formation of a comprehensive player behavioral feature dataset. Subsequently, we harnessed six distinct machine learning models to prognosticate match results in the domain of wargaming using this dataset, achieving a peak prediction accuracy of 96.11%. The primary emphasis lies in the identification of prevalent determinants contributing to player triumphs in wargaming. To this end, we conducted an attribution analysis to ascertain the significance of diverse macro and micro features. Guided by the importance of these features, we propose a method for evaluating player performance. This methodology can be instrumental in scrutinizing disparate player wargaming styles, dissecting customary strategic behaviors that lead to player victories, and assisting wargame designers in crafting AI agents capable of adapting to a spectrum of human player behaviors. Consequently, this study offers substantial insights for the advancement of research in the realm of human-AI hybrid gameplay.

With the development of society, intelligent games have gradually become a hot research field. This article proposes an algorithm that combines the multiattribute decision-making and reinforcement learning methods to apply to multiagents’ decision-making for wargaming artificial intelligence (AI).This algorithm solves the problem of the agent's low rate of winning against specific rules and its inability to quickly converge during intelligent wargame training. At the same time, a multiattribute decision-making method based on the entropy–weight method was proposed to obtain the normalized weighting for each attribute that feeds into a deep reinforcement learning model. A simulation experiment confirms that the real-number multiattribute decision-making-proximal policy optimization (PPO) algorithm of multiattribute decision-making combined with reinforcement learning presented in this article is significantly more intelligent than the pure reinforcement learning algorithm.

Presently, intelligent games have emerged as a substantial research area. Nonetheless, the slow convergence of intelligent wargame training and the low success rates of agents against specific rules present challenges. In this article, we propose a game confrontation algorithm combining the multiple attribute decision making (MADM) approach from management science and reinforcement learning (RL) technology. This integration enables us to combine the strengths of both approaches and addresses the above issues effectively. This study conducts experiments using the algorithm that integrates MADM and RL techniques to gather confrontation data from the red and blue sides within the winning-first wargame platform. The data is then analyzed using the weight calculation method of intuitionistic fuzzy numbers to determine each intelligent opponent agent's threat level from the perspective of MADM. The threat level calculated by MADM is used to construct the reward function for the red side. The simulation results demonstrate that the algorithm combining MADM and RL proposed in this study outperforms classical RL algorithms regarding intelligence. This approach effectively addresses issues, such as the convergence difficulty, caused by random initialization and the sparse rewards for agent neural networks in wargame environments with large maps. Combining the MADM method from management with the RL algorithm in control can lead to cross-disciplinary innovation in academic fields, which provides innovative research values for intelligent wargame design and RL algorithm improvements.

The application of artificial intelligence (AI) in games has been significantly developed and attracted much attention over the past few years. This article not only leverages the reinforcement learning multiagent deep deterministic policy gradient algorithm to realize the dynamic decision-making of game AI but also creatively incorporates deep learning and natural language processing technologies in the wargame field to transform game context situation maps into textual suggestions in wargame confrontation. In this article, we effectively integrate reinforcement learning technologies, deep learning technologies, and natural language processing technologies to generalize the semantic text output at state-of-the-art accuracy, which plays an important role in human understanding of game AI behavior. The experimental results are promising and can be used to verify the feasibility, accuracy, and performance of our proposed model in extensive simulations against benchmarking methods.