Bias, Discrimination, and Fairness

The MD4SG Bias, Discrimination, and Fairness working group examines the various social ramifications of technological systems that are deployed in real-world settings. We are especially concerned with the perpetuation and normalization of socially unjust practices that occur when seemingly objective tools embed and operate from biased understandings of the world. Drawing on a variety of academic disciplines, we turn our focus inward to assess how various technical approaches can and cannot be more thoughtfully designed and implemented so to prevent the further marginalization of our most vulnerable communities.

Our group consists of bi-weekly seminar-style talks of recent work, project proposals, and subarea overviews. Below are some examples from Spring 2020:

Recent seminar talks #

• The benefits of diversity in information diffusion(Ana-Andreea Stoica)
• Fair ML and the Domain Adaptation Problem (Alexander Ritchie)
• Advertising for a demographically representative cohort (Lodewijk Gelauff)
• Privacy Legislation Amendments to PIPEDA (Canadian Privacy Law) to accommodate impacts of AI (Abhishek Gupta)
• Fairness via Separation of Powers (Ethan Cowan)
• How Do We Define “Fair” Responsibly? (Manuel Sabin)
• Fairness questions in data elicitation and aggregation (Jessie Finocchiaro)
• Paradoxes in Fair Machine Learning (Anson Kahng)
• Homophily and Monopoly-Monopsony Power in Data and Job Markets (Sara C. Kingsley)
• Ethics companion to textbook: Why? What? Can you join us? (Augustin Chaintreau)
• Overview of bias, discrimination and fairness in ML and online labor markets (Faidra Monachou, Manish Raghavan)

Projects #

• Fairness and Discrimination in Machine Learning and Mechanism Design. Appeared at AI4SG workshop at HCRCS July 2020 (link).
• Problems of scale in content moderation
• Exploring the space of long-term fairness in homophily-driven systems

Working Group Organizers #

Working Group Members #