Grant: Alliance Grants

Recent advances in AI, particularly foundation models are playing an increasingly important role and transforming the way we develop software. Today, many organizations in Canada (and internationally) are integrating these foundation models (e.g., large language models) in their daily software development. However, integrating AI and building AI-based software systems poses many challenges that require a comprehensive investigation on the benefits and challenges of employing AI in the development of AI-based software systems. Hence, the goal of this partnership between Concordia University and the National Bank of Canada is to improve the development of trustworthy AI-based software systems using AI. We address this goal by focusing on two aspects: the processes and teams that develop AI-based systems and the building of the AI-based software systems themselves. The anticipated outcomes of the proposed program are many-fold. First, novel metrics to determine the effectiveness of using foundational models in the development processes of AI-based systems, empirical findings on the effectiveness of using foundation models in the development of AI-based software systems and techniques to improve the use of foundation models in the development of AI-based software systems will be developed. Furthermore, analytics to determine the trustworthiness of software artifacts from foundation models, techniques to effectively evolve software artifacts from foundation models, techniques to map and address production issues to source code and tools that operationalize the developed techniques will also be developed. The outcomes of the proposed research will lead to significant advancement in the development of AI-based systems using foundation models.