Bias and Fairness Across the Machine Learning Pipeline

Despite the common perception of AI systems as objective and data-driven, each stage of ML development reflects human choices and social contexts that can encode or amplify inequities. The talk discusses key sources of harm and different approaches to understanding them, showing that fairness in ML goes beyond technical fixes and involves broader ethical and social considerations. Finally, it briefly presents one of our recent works on addressing intersectional bias in ML systems.

Speaker's brief bio:

Amir Payberah is an Associate Professor of Computer Science at KTH Royal Institute of Technology, where he leads the co-liberative computing research group. His work explores equity and justice in AI, aiming to integrate critical awareness, inclusivity, and societal good into computing research and education. Dr. Payberah is a guest of the Distributed Computing and Systems research group and will be giving a guest lecture in the TDA596 Distributed Systems course.