This paper reports on the use of a structured argument to evaluate safety risk associated with the use of an AI/ML enabled system to monitor telemetry communicated to a ground station by a robotic arm being used in human spaceflight. In addition to addressing questions about the performance of the AI/ML enabled system (e.g., recall rate, false positives), the structured argument addresses human factors considerations such as the risk that operators might become complacent by trusting the AI-functionality to monitor operational safety. Rather than just using a structured argument to document the outcome of a safety assurance process, the evaluation approach is being driven by a structured argument concurrently with development of the system. Taking an approach based on a GSN variant, namely, Eliminative Argumentation, the use of defeaters (e.g., “unless the operator becomes de-sensitized to AI/ML generated reports of anomalies due to a high rate of false positives”) is an effective means of guiding systematic evaluation of a new technology to better understand its impact on an operational process. A similar approach could be used in other technical domains such as medical devices, critical infra-structure and plant automation that are rapidly integrating AI/ML into safety-critical technology.