Failure Modes and Effects Analysis of AI-Enabled Technologies for Aeronautical and Astronautical Applications

Abstract

AI technologies (AIT) have the potential to transform business processes by increasing productivity and reducing costs, but their adoption may also introduce unintended consequences. The overarching objective of this study is to examine the key promises and perils of AIT in the aerospace industry and to provide recommendations for safer AI use across the aerospace ecosystem, both airborne and ground-based. To this end, we adapt the traditional failure modes and effects analysis (FMEA) method to the aerospace domain—encompassing aeronautics and astronautics—and designate the resulting approach as the AIT-FMEA framework. Applying AIT-FMEA, we identify 23 AIT-related failure modes (FMs), characterize their potential effects, and propose targeted risk-mitigation strategies. The FMs are grouped into five categories: Technical [T], Social/Societal [S/S], Institutional/Organizational [I/O], Environmental [E], and Political [P]. Of the identified FMs, 43.48% fall under [T], 21.74% under [S/S], 17.39% under [I/O], 13.04% under [P], and 4.35% under [E]. We map these FMs to relevant aerospace standards, including ARP4761A, DO-178C, DO-326B, and DO-356A. A two-factor (4×5) risk model indicates that, if unmitigated, these FMs could pose medium to critical risks, exceeding an acceptability threshold defined by improbable or remote likelihood combined with low severity. Addressing these FMs is essential to realize the benefits of AIT while avoiding undue risk. The study’s insights are intended to support AI policymakers, aerospace standards bodies, and AIT developers in prioritizing verification and validation (V&V) and implementing other proactive measures to manage AIT-related risks.