Understand the conceptual field of AI transparency, including terminology, history, and primary ethical considerations.
Examine frameworks for explainable AI (xAI)
Reading
Consider While Reading:
As you read Christian's chapter, please pay attention to how the author points toward the connections between transparency and accountability. Notice also the tensions the chapter raises: transparency can increase trust and fairness, but it can also overwhelm users with information or be strategically used to obscure responsibility. As you go through the text, keep asking: Transparency for whom? Transparency about what? And transparency to what end? These questions will help you connect the chapter's insights to broader debates in AI ethics about fairness, power, and the critical benefits, but also the limits of technical-only solutions to social problems.
Research Connection
This article by Anthropic (the company that creates Claude) introduces a new interpretability method—circuit tracing—that maps how a language model transforms a prompt into an answer, step-by-step, revealing intermediate “plans” and concept representations rather than just the final output.
Additional Readings (Optional)
Cynthia Rudin: Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges.
Parts of this paper may be too technical; the Principles portion (first nine pages) introduces an important perspective on XAI.
Felzmann et al. - Towards Transparency by Design for Artificial Intelligence