Integrators at War: Mediating in AI-assisted Resort-to-Force Decisions

The integration of AI systems into the military domain is changing the way war-related decisions are made. It binds together three disparate groups of actors - developers, integrators, users - and creates a relationship between these groups and the machine, embedded in the (pre-)existing organisational and system structures. In this article, we focus on the important, but often neglected, group of integrators within such a sociotechnical system. In complex human-machine configurations, integrators carry responsibility for linking the disparate groups of developers and users in the political and military system. To act as the mediating group requires a deep understanding of the other groups’ activities, perspectives and norms. We thus ask which challenges and shortcomings emerge from integrating AI systems into resort-to-force (RTF) decision-making processes, and how to address them. To answer this, we proceed in three steps. First, we conceptualise the relationship between different groups of actors and AI systems as a sociotechnical system. Second, we identify challenges within such systems for human-machine teaming in RTF decisions. We focus on challenges that arise a) from the technology itself, b) from the integrators’ role in the sociotechnical system, c) from the human-machine interaction. Third, we provide policy recommendations to address these shortcomings when integrating AI systems into RTF decision-making structures.

💡 Research Summary

The paper “Integrators at War: Mediating in AI‑assisted Resort‑to‑Force Decisions” examines the often‑overlooked role of integrators—actors who bridge AI developers and military‑political users—in the deployment of artificial‑intelligence systems that support high‑level resort‑to‑force (RTF) decision‑making. The authors frame the entire ecosystem as a socio‑technical system composed of three distinct groups (developers, integrators, users) and the AI machine itself, all embedded within existing organisational structures.

First, the authors contextualise the problem historically, comparing the integration of aircraft carriers and cryptographic tools into military operations. These examples illustrate how successful integration can transform warfare, while poor integration can cause catastrophic failures, underscoring the pivotal influence of the “sandwich” actors who must align disparate technical, operational, and cultural requirements.

Next, the paper distinguishes AI integration from earlier technologies. Unlike cryptography, where the goal (secure communication) is clear and failures are observable, AI‑RTF systems often lack a well‑defined objective, produce opaque outputs, and operate at varying levels of automation. Consequently, integrators face novel challenges: (1) technological uncertainty (bias, opacity, shifting responsibility), (2) organisational ambiguity (translating developer jargon into military doctrine, reshaping decision‑making processes), and (3) human‑machine interaction issues (designing explainable interfaces, maintaining trust, ensuring a robust human‑in‑the‑loop).

The authors propose a “technology‑integrator‑HMI” triple‑loop model to map these interdependencies and identify risk vectors. For each loop they outline mitigation strategies: rigorous pre‑deployment testing, continuous monitoring, explainability tools, dedicated translation and coordination functions, and training programmes that teach users how to interrogate AI recommendations critically.

Building on prior work by Chiodo et al. (2024a), the paper adapts developer‑focused recommendations into three pillars for responsible AI integration: (1) Accountability – clear attribution of decisions across the chain; (2) Transparency – traceable data pipelines and model documentation; (3) Capability – specialised curricula and certification for integrators. These pillars are positioned as complementary to NATO’s AI principles and the EU AI Act, filling a regulatory gap concerning the “integration” phase.

Finally, the paper offers concrete policy recommendations: (i) institutionalise integrator roles within defence ministries, (ii) create accredited training pathways that blend systems engineering, ethics, and military doctrine, (iii) embed responsibility‑tracking mechanisms (audit logs, decision‑records) to enable post‑hoc review, and (iv) foster a culture of interdisciplinary collaboration to reduce silos between developers and end‑users. The authors argue that making integrators visible and empowering them with the right tools and authority is essential to prevent AI‑driven miscalculations that could trigger unnecessary conflict or exacerbate civilian harm.

In sum, the article highlights that the success or failure of AI‑assisted RTF decisions hinges less on the underlying algorithms and more on the quality of the mediating work performed by integrators, calling for systematic investment in their expertise, governance structures, and accountability frameworks.