Dynamic Autonomy Management in Human-AI Command and Control for Autonomous Weapons Systems

The accelerating deployment of autonomous weapons systems (AWS) in military operations has created an urgent need for empirically validated frameworks governing the dynamic allocation of decision authority between human commanders and artificial intelligence systems. This dissertation developed, tested, and validated a Dynamic Autonomy Management (DAM) framework for human-AI command and control (C2) in autonomous weapons employment through a four-phase sequential mixed-methods design. Phase 1 applied grounded theory analysis to an 84-document corpus of policy directives, government reports, and international legal instruments, identifying Autonomy Governance as the core category with the highest centrality score among eight emergent categories. Phase 2 employed agent-based computational modeling with 13,500 Monte Carlo iterations across three C2 architectures and three threat conditions, quantifying the fundamental speed–accountability tradeoff: human-in-the-loop (HITL) maintained 97.8% accountability chain integrity but with 8.51-second mean response latency, while human-over-the-loop (HOVL) achieved 1.20-second latency at the cost of reduced accountability (68.2%). Phase 3 conducted simulation-based experimentation with 118 participants in a 3 × 3 factorial design, confirming large effects of autonomy level on response time (η²p = .73) and a trust–accuracy paradox wherein higher-autonomy systems produced better objective performance but lower operator trust. Phase 4 convened expert tabletop exercises with 18 defense professionals who rated the DAM framework positively across all five evaluation criteria, with Decision Traceability receiving the highest rating (M = 5.83, SD = 0.62) and Scalability the lowest (M = 4.72, SD = 1.18). Human-on-the-loop (HOTL) architecture consistently emerged across all four phases as the optimal default configuration, balancing operational tempo with meaningful human control. The DAM framework provides an empirically grounded governance architecture for dynamic transitions of decision authority between human operators and autonomous weapons systems, directly informing Joint Chiefs of Staff doctrine development, DoDD 3000.09 implementation, and international autonomous weapons governance.