On the Road to Designing Responsible AI Systems in Military Cyber Operations

Abstract

Military cyber operations are increasingly integrating or relying to a specific degree on AI-based systems in one or more moments of their phases where stakeholders are involved. Although the planning and execution of such operations are complex and well-thought processes that take place in silence and with high velocity, their implications and consequences could be experienced not only by their targeted entities, but also by other collateral friendly, non-friendly, or neutral ones. This calls for a broader military-technical and socio-ethical approach when building, conducting, and assessing military Cyber Operations to make sure that the aspects and factors considered and the choices and decisions made in these phases are fair, transparent, and accountable for the stakeholders involved in these processes and the ones impacted by their actions and largely, the society. This resonates with facts currently tackled in the area of Responsible AI, an upcoming critical research area in the AI field that is scarcely present in the ongoing discourses, research, and applications in the military cyber domain. On this matter, this research aims to define and analyse Responsible AI in the context of cyber military operations with the intention of further bringing important aspects to both academic and practitioner communities involved in building and/or conducting such operations. It does that by considering a transdisciplinary approach and concrete examples captured in different phases of their life cycle. Accordingly, a definition is advanced, the components and entities involved in building responsible intelligent systems are analysed, and further challenges, solutions, and future research lines are discussed. Hence, this would allow the agents involved to understand what should be done, what they are allowed to do, and further propose and build corresponding strategies, programs, and solutions e.g., education, modelling and simulation for properly tackling, building, and applying responsible intelligent systems in the military cyber domain.