Systems Engineering-Driven AI Assurance and Trustworthiness

Artificial intelligence (AI) across the engineering domain has led to an increased discussion about the importance of creating, maintaining, and governing AI, and AI assurance has been discussed at multiple forums at times without specific steps to make the process actionable. This work proposes that AI assurance can be addressed through five intertwined components: ethics, transparency, compliance, safety, and certification. AI assurance's importance comes from an increase in complex algorithms that learn from data corresponding to several studies and massive data banks that may exhibit bias and variance. In addition, the surge in AI has given rise to opportunities for the community to use systems engineering for AI assurance. This work develops a method to connect continuous integration/continuous deployment (CI/CD) pipelines and subject matter experts to address AI assurance in a systematic and semi-automated approach. The work demonstrates the method through a use case involving a relevant dataset.