Responsible & Explainable AI

Course Code: Y1C3
ECTS Credits: 4.0

Course Description

This course offers a practical and reflective introduction to responsible & explainable AI (XAI), focusing on the identification of bias and the importance of fairness, transparency, and interpretability in AI systems. You will begin by performing an exploratory data analysis to uncover hidden biases in an image dataset. After building and training your own image classification model, you will explore how to make your model more transparent and understandable using a variety of XAI techniques.

The course also supports the development of essential academic and ethical reasoning skills. Through a Socratic dialogue you will examine key concepts such as fairness, bias, equity, equality, and worldview. These concepts will be explored in relation to both personal perspectives and shared societal values.

As part of the learning process, you will connect different types of bias to the phases of the data science process (CRISP-DM), and consider how decisions made at each stage can impact outcomes. You will explore different ways of thinking about fairness, both at the individual and group level, and reflect on the tensions that can arise between competing goals.

In the second half of the course, you will be introduced to the emerging field of explainable AI. You will learn to apply methods that help reveal how AI systems arrive at their predictions, and reflect on what makes an explanation useful or meaningful in different real-world contexts. During a group presentation, you will also consider the trade-offs between accuracy and interpretability, and propose use cases where one might be prioritized over the other.

By the end of the course, you will be able to identify and analyze bias in datasets, apply XAI techniques to improve AI model interpretability, and critically evaluate the ethical dimensions of AI systems. This course provides a strong foundation for building AI that is not only technically robust, but also socially responsible.

Course Content

  • Define key terms such as fairness, equity, equality, implicit and explicit bias, and worldview
  • Identify and analyze bias in image datasets and connect findings to stages in the data science process (CRISP-DM)
  • Explore personal and collective worldviews and their impact on how fairness is understood and applied
  • Compare different approaches to fairness and consider how individual and group goals may conflict
  • Investigate ethical issues in AI through structured research and engage in Socratic dialogue to deepen critical thinking, ethical reasoning, and collaborative understanding
  • Learn the importance of interpretability and transparency in AI systems
  • Apply XAI methods that help clarify how models make decisions
  • Reflect on the characteristics of a good explanation and when to prioritize interpretability over accuracy
  • Communicate insights clearly through in-class presentations and discussions

Prerequisites

  • Introduction to Programming with Python (Y1A1)
  • Introduction to AI (Y1A3)
  • Introduction to Machine Learning (Y1B1)

Course Coordinator(s)