The learning module is designed in the spirit of open and distance education, and can be followed at each individual’s own pace. To provide an introduction to the ongoing work in FATE, we have structured the module into five thematic units:
We begin with the broad topic of AI Ethics, which addresses ethical concerns in any artificial intelligence (AI) system. As a field, AI Ethics aims to provide technical solutions to promoting fairness, accountability and transparency in AI systems, and in this unit, we examine some of the key challenges.
Our second unit focuses on the issues posed by the rise of machine learning and data-driven AI. This unit provides an overview of the sources of algorithmic biases (e.g., data, algorithms, users), as well as specific solutions highlighted in the research to date, aimed at promoting fairness.
While Unit 2 focuses on the identification of algorithmic bias and the promotion of fairness, in Unit 3 we consider the importance of transparency. In particular, this unit provides a survey on techniques for promoting explainability, such that users can more easily interpret an algorithmic system’s output, and use it effectively and appropriately.
In Unit 4 we examine the very important, practical issue of FATE awareness. Specifically, we examine the views of various stakeholders of data-driven AI systems (e.g., developers, end-users), in an effort to understand how we might raise their awareness of FATE and their own role in promoting more ethical development and use of AI systems.
In the final unit of the module, we examine FATE issues in the context of particular application areas and domains. Specifically, we consider applications such as web search engines, algorithmic curation and filtering mechanisms such as those used in social media platforms, and computer vision (image tagging) in an effort to understand how FATE might impact our everyday interactions with modern information systems.
Each unit begins with a brief introduction to the topic, as well as an overview of the specific learning goals. The material for each unit consists of educational videos - most of which have been created through CyCAT workshops and seminars - along with the lecturers’ slides, as well as a supporting bibliography (i.e., academic articles on the topic). At the end of each thematic unit, there is also a quiz, which can be used to gauge one’s understanding of the material covered. Feedback is immediately provided via Google Forms, once the responses are submitted.
As a self-taught module, learners may of course direct their own study in any way they like. However, we would estimate that a rigorous treatment of each unit would take somewhere between 10 and 15 hours.
For the last four units of the module, we have provided one or two activities, which enable learners to apply the respective concepts and techniques of the unit. These are exercises that we have used with our own students and/or in the context of the CyCAT workshops and winter school conducted during our project.