Trusting Co-Intelligence: Method and Case Studies on how to manage Human-AI decision making

Artificial Intelligence (AI) is revolutionizing the way we work, necessitating new collaborative models that blend human and AI capabilities. Despite its potential, effective AI deployment is hampered by trust issues. This thesis introduces a novel framework to manage trust in human-AI decision-making, focusing on four key factors: transparency, accountability, similarity, and performance. Grounded in the vast literature on trust (spanning from psychology, computer science and management) and tested across various projects in the education sector and a predictive maintenance context, the framework uses questionnaires and brainstorming sessions to identify enablers and barriers to AI trust, as prerequisite of AI correct usage in projects. The study emphasizes the importance of addressing both technical and social dimensions of trust, providing concrete managerial tools to enhance AI acceptance. The flexible framework can be applied proactively or reactively, making it suitable for different stages of AI projects. Future research should systematically analyse data to identify patterns and guidelines for improving AI trust and deployment strategies.