Bias and Fairness
Lecture
This week's lecture covers the concepts of Bias and Fairness.
Slides:
Document
Transcript:
Reading
Required - Fairness Measures
"The (Im)possibility of Fairness: Different Value Systems Require Different Mechanisms For Fair Decision Making"
https://dl-acm-org.ezproxy.is.ed.ac.uk/doi/10.1145/3433949
(old link: https://cacm.acm.org/magazines/2021/4/251365-the-impossibility-of-fairness/)
This short (but quite concept-dense) article contrasts different measures of fairness, and what assumptions they are making about the problems they're applied to.
Optional - Is Bias even the right problem?
"The Seductive Diversion of ‘Solving’ Bias in Artificial Intelligence"
This article argues that focusing on bias as a core problem of AI actually allows other, worse, problems to be missed.
Optional - Practical Example in Diversity
"Diversity and Inclusion Metrics in Subset Selection"
https://dl.acm.org/doi/pdf/10.1145/3375627.3375832
This paper gives an example of a more grounded practical example of quantifying fairness measures.
Optional - Challenges in getting data for Fairness checks
""What We Can't Measure, We Can't Understand": Challenges to Demographic Data Procurement in the Pursuit of Fairness"
https://arxiv.org/abs/2011.02282
This paper talks about the difficulty of getting the nceessary data for performing fairness checks, when often this is at odds with unbiased data collection in the first place.