Synthesized Command and Control: Guiding AI in Warfighting
- •Authors propose SYNTHComm to embed human intent directly into military AI software architectures.
- •Current military AI decision-making models face challenges balancing human oversight with operational speed.
- •SYNTHComm uses frequency-based governance to distribute authority across system logic, rather than episodic intervention.
The U.S. military faces a challenge in managing increasingly autonomous AI systems, where traditional human-in-the-loop oversight often proves too slow for the tempo of modern conflict. The article argues that waiting for human approval on individual strikes is inefficient and can constrain AI performance. Instead, authors Dr. Elise Annett and Dr. James Giordano propose a framework called Synthesized Command and Control (SYNTHComm) to move governance from external, episodic human intervention to persistent, internal system logic.
SYNTHComm encodes commander intent, policy constraints, and operational goals directly into the software as weighted functions and structured constraints. This approach treats agentic behavior (systems capable of autonomous action) as a frequency spectrum: lower-frequency components manage long-term mission objectives and policy, while higher-frequency components handle real-time adaptations to the battlefield. By encoding this authority within the architecture, the system maintains alignment with military objectives without needing constant manual verification.