Air Force AI Slashes Targeting Time to Seconds, but Reliability Gaps Persist
DASH-2 experiments show AI generating targeting options in eight seconds versus 16 minutes for humans, though 'non-viable' results mandate human oversight.
The Department of the Air Force is accelerating the integration of artificial intelligence into tactical command and control to shrink the window between detecting a threat and engaging it. Recent experiments demonstrate that while AI can process data at speeds unattainable by humans, it still lacks the operational judgment required to function autonomously.
During the "Decision Advantage Sprint for Human-Machine Teaming" (DASH-2) experiment, AI-enabled microservices generated courses of action (COAs) for weapon-to-target matching in approximately eight seconds. In contrast, human operators required 16 minutes to produce their options. The AI also demonstrated a higher volume of output, generating 10 different COAs compared to just three produced by the human team. These tests were conducted at the Shadow Operations Center-Nellis (SHOC-N), the U.S. Air Force's premier tactical command and control battle lab.
The Path to Joint All-Domain Command and Control
The DASH sprints are part of the broader Advanced Battle Management System (ABMS), which serves as the Air Force's primary contribution to the Pentagon's Joint All-Domain Command and Control (JADC2) concept. The goal of ABMS is to move away from disparate, "stovepiped" systems toward a cohesive enterprise battle network. By involving operational warfighters, industry partners, and Space Force representatives, the Department of the Air Force is attempting to transition human-machine teaming from a theoretical framework into an operational reality.
The Risk of Automation
Despite the leap in speed, the experiments highlighted a critical vulnerability: the AI frequently produced non-viable options. Maj. Gen. Robert Claude, a Space Force representative to the ABMS Cross-Functional Team, noted that while the AI was more timely and generated more options, those options were not necessarily viable. This gap in reliability underscores the danger of over-reliance on automation in high-stakes environments, where a hallucinated or illogical tactical plan could lead to mission failure.
Col. Jonathan Zall, ABMS Capability Integration chief, stated that DASH-2 proved human-machine teaming is no longer theoretical, suggesting that the future of decision advantage lies in fusing operator judgment with AI speed.
Next Steps for Tactical AI
The results validate the military's current strategy of maintaining a "human-in-the-loop" to ensure safety and viability. Future iterations of the DASH experiments will likely focus on refining the AI's ability to account for real-world constraints and reducing the frequency of non-viable COAs. The Air Force continues to prototype these microservices to determine how to best balance the raw processing power of AI with the nuanced experience of professional warfighters in joint and coalition operations.