Designing Responsible AI Solutions: Exploring AI Governance & Ethics through a KPMG Case Study

The present thesis, titled Designing Responsible AI Solutions: Exploring AI Governance & Ethics through a KPMG Case Study, takes an in-depth look at the complex field of governance and ethics with respect to Artificial Intelligence (AI). This is an extremely timely topic as AI continues to spread across multiple sectors and aspects of modern life; addressing the challenges and opportunities emerging from the wide adoption of AI requires ongoing dialogue, appropriate regulation, and critical reflection on the ethical implications of rapidly evolving AI. The goal of this research is dual. First, it aims to provide a comprehensive overview of the topic of AI and the governance and ethics aspects of AI, both at the theoretical and regulatory levels. Second, it aims to identify and propose concrete solutions to contribute significantly to promoting responsible governance and ethics in the use of AI at the corporate level. The paper analyses a real-life case study derived from the internship experience at KPMG, which helps to capture the governance aspects and ethical considerations to be taken into account when an AI system is created. This analysis is critical to devising a possible solution to handle governance and ethics issues; in fact, an AI Ethics Checklist is proposed that every employee can refer to when working with AI systems. To ensure its validity and effectiveness, the Delphi method, an evaluation procedure involving experts providing anonymous answers on a complex topic, was used, allowing a consensus to be reached; the relevance and usefulness of this checklist will later be evaluated through further measurement when it is implemented in KPMG's corporate environment. The AI Ethics Checklist is a fundamental tool for promoting responsible governance and ethics in the use of Artificial Intelligence, resolving a gap often found within government authorities, which tend to provide abstract guidelines that are difficult to apply.