Grupo Alarcos · Conference paper · 2027

Towards a Maturity Model for AI Software Processes with International Standards

Communications in Computer and Information Science · 2027

Artificial intelligence (AI) has become a key driver of innovation across multiple sectors, increasing the need to ensure not only functional performance but also the quality and reliability of AI systems. New regulations directly focused on the field of AI are emerging, such as the EU AI Act, along with new international standards such as ISO/IEC 5338:AI system life cycle processes, which attempt to address the importance of the quality of AI development processes; however these lack a company-oriented approach. This paper proposes the Artificial Intelligence Software Maturity Model (MMSIA), a standardized framework designed to assess and improve the capability and maturity of AI system development processes. The MMSIA integrates the life cycle processes of ISO/IEC 5338 with the assessment framework of the ISO/IEC 33000 family, enabling organizations (especially SMEs) focused on AI system development to assess and continuously improve their AI development practices. The model defines a total of 24 processes required in the AI development lifecycle, divided into five maturity levels, enabling a structured approach that promotes advancement toward the development of reliable and trustworthy AI systems.

View on the group website DOI: 10.1007/978-3-032-37353-3_4

On the same topic