Learning Unit 4: Transparency
In this unit, we establish transparency as a core value of trustworthy AI, requiring systems to be traceable, explainable, and clearly signal AI interaction. Students learn about a human-centered approach to transparency, defining the context and justifying and selecting methods (white-box models vs. post-hoc XAI like SHAP/LIME). A real-world case illustrates the societal harm of not ensuring transpareny in algorithmic systems.
Exercises
Task 1 - Transparency
This unit introduces the concept of white-box and black-box models and which tools to use to help interpretation and model assessment.
