This collection focuses on the concept of interpretability and transparency in artificial intelligence across various domains. It covers applications in healthcare, finance, and network management, detailing methods such as Item Response Theory and SHAP for analyzing algorithms and making predictions comprehensible. Additionally, it examines the ethical implications of AI decision-making, emphasizing the importance of fairness, accountability, and collaboration in tool development. The content underscores the integration of AI with human-centric approaches to enhance trust and effectiveness in diverse fields.
Perché l’AI si perde quando cambiamo strada? Problemi di generalizzazione nella previsione della mobilità urbana