This module is refreshed on an annual basis, with the aim of providing students with exposure to the most current topics in interactive technologies. The current topic, Fairness, Accountability, Transparency and Ethics (FATE) in AI Systems, focuses on the interaction with algorithmic processes - including the key technical, ethical, and social issues that arise. HCI researchers and practitioners will increasingly need to understand FATE issues and how to address them.

Intelligent systems have become ubiquitous across sectors of society. While their adoption has brought about many positive changes in our everyday lives, they have also introduced several social and ethical challenges. In particular, there is much concern surrounding the use of data-driven, machine learning development techniques that often result in systems that are essentially “black boxes” – neither the developer nor the user is positioned to fully understand or interpret their behaviors. 

Furthermore, industrial trends have also played a role in the proliferation of opaque systems in citizens’ lives. In the “Algorithm Economy” companies sell not only products but also “algorithms as a service.” Thus, developers are increasingly incorporating proprietary cognitive components (e.g., image recognition, natural language processing, or search functionalities via Cognitive Services) into their own software. However, with increasing legislation (e.g., the EU’s General Data Protection Regulation and eventually, the AI Act) and industry standards that promote responsibility and accountability in intelligent systems, there is the potential that developers will be held accountable for the behaviors and consequences of the systems they build.

This module aims to teach students how to identify and analyze the potential social and ethical implications of algorithmic systems and services. It adopts a practical approach; students will become familiar with the state-of-the-art research and techniques related to “FATE” in algorithmic systems, and in particular, the detection and mitigation of algorithmic bias and the promotion of transparency and interpretability (explainable systems) through a series of lectures, and by getting hands-on experience with approaches and tools for analyzing machine behavior for the purpose of mitigating ethical and social concerns.

Module content

1. Ethical AI
2. AI systems and user interaction
3. User trust in AI systems 
4. Personal privacy 
5. Discrimination and bias
6. How machines learn to discriminate 
7. Bias detection - Auditing
8. Promotion of fairness - Background 
9. Fairness perception 
10. The limits of transparency
11. Explainable systems – Background
12. Application areas