Exploring the Role of Explainable AI in the Development and Qualification of Aircraft Quality Assurance Processes: A Case Study

Quality assurance using non-destructive inspection (NDI) is an essential domain in aircraft manufacturing that can benefit greatly from the latest developments in artificial intelligence (AI). A barrier to the use of AI in this domain is the lack of a qualification philosophy, based on the lack of trustworthiness. In our article, we explore the role of explainable artificial intelligence (xAI) as a possible technology building block in the later qualification philosophy of AI systems in NDI. We therefore select a specific case study, where a computer vision-based model has been previously trained to predict material defects from process monitoring data in additive manufacturing. We show the benefits xAI can bring to the development and qualifiability of NDI using AI, and also highlight limitations and current gaps of xAI in the presented context. With the dynamic changes in the regulatory landscape, there is now an opportunity for xAI to demonstrate its ability to support the validation of AI systems in aircraft manufacturing and therefore proving its value to be incorporated into the qualification processes.