Synergistic Optimization of Human and Artificial Agents in Labor Resource Management in Hybrid Sociotechnical Systems

Introduction. The rapid development of artificial intelligence (AI) technologies and the associated transformation of organizational processes are radically changing traditional approaches to human resource management (HRM). Classical optimization methods often prove insufficient to solve new dynamic problems faced by modern organizations and do not take into account the growing complexity of effective application of AI. The emergence of hybrid human + AI teams, automation of routine TR processes, the emergence of career trajectory recommendation systems, predictive analytics of fluidity and affective computing create a fundamentally new field of interaction in which the traditional dichotomy "manager – subordinate" gives way to the synergistic interaction of multiple agents (human and artificial) with different cognitive architectures, goals and constraints. The article proposes concepts and methodologies for synergistic optimization of human and artificial agents in the field of labor resource management of hybrid organizations, allowing to improve the work of the organization. The purpose of the work is to study modern approaches to optimizing labor resources based on AI agents, critically analyze existing models, formulate new mathematical statements of tasks and determine directions for further research. The development of algorithms for optimizing "human-AI" interaction allows you to reduce the costs of personnel management, while maintaining or increasing productivity. Thus, the work clearly outlines the problem of the lack of comprehensive models of "Human + AI" interaction, taking into account the degradation of both parties. Based on this, an own integrated model is proposed, aimed at optimizing computational and human resources. The results of the study show that the integration of autonomous and cognitive AI agents into labor resource management systems forms a new paradigm of organizational efficiency, which is significantly different from traditional TR approaches. The focus of change is not the replacement of human labor, but the construction of hybrid socio-technical systems, where human abilities are supplemented by the computational capabilities of AI. The mathematical models presented in the article show that the combination of discrete optimization methods, game theory, queueing models, multi-agent systems and machine learning algorithms creates a toolkit capable of providing balanced load management, preventing personnel fatigue and preventing technical degradation of AI agents. Conclusions. The proposed formal statements of optimization problems, in particular task assignment models, two-level hierarchical management structures and behavioral models in the form of Markov processes, make it possible to unify the processes of interaction between people and agents. Analysis of the influence of the orchestrator showed that the introduction of a super-level coordinator transforms the system into a controlled dynamic architecture, ensuring global consistency and stability even under conditions of stochasticity and high variability of tasks. At the same time, the study of behavior trees proves that the formalization of local behavior of agents significantly increases the predictability and reliability of the system, but requires careful synchronization with the orchestrator policies to avoid loss of flexibility.