AI Accountability Framework

In recent years, the adoption of Artificial Intelligence (AI) in business decision-making has accelerated, promising gains in operational efficiency and competitive advantage. AI is a branch of computer science, focused on developing systems that perform tasks traditionally requiring human intelligence, such as decision making, reasoning, natural language processing and predictive analytics. This thesis focuses specifically on AI applications leveraging classic statistical methods and machine learning models such as used for churn prediction, credit scoring, customer segmentation, excluding generative AI approaches. The increasing complexity of AI models and regulatory demands pose significant operational, compliance and reputational risks. While numerous frameworks outline responsible AI principles, practical guidance for their implementation remains limited. To address this, the thesis introduces a governance framework based on Design Science Research methodology, providing practical tools for enhanced AI oversight at organizations. Aligned with EU regulations requiring comprehensive model documentation and registration, the framework emphasizes maintaining a centralized model registry with audit trails for transparency and ease of compliance. Designed for adaptability across industries, it offers practical solutions using widely available software such as Microsoft Excel, ensuring accessibility for organizations with limited resources. The primary goal of this thesis is to bridge theoretical guidelines with practical applications, enabling organizations to responsibly manage AI deployment, improve governance, compliance and promote sustainable practices.