Module 2: Bias and types of harms in data

Biases are an intrinsic part of human behaviour, and are reflected in human-generated data. Not all types of biases are undesired, but many of them can produce harms. In this second module we will explore how biases can be embedded in data and learning systems. After, we will see how certain biases can be detected and mitigated.

Readings

Videos:

• Joy Buolamwini (2016): How I’m fighting bias in algorithms — 9 minutes

Printed materials:

• Mortiz Hardt (2014): How Big Data is Unfair

• Geomblog (2019): On centering, solutionism, justice and (un)fairness.

• Mirror mirror. Reflections on Quantitative Fairness

• American College of Cardiology - Speaking from the heart: Could your voice reveal your heart health?

Debate: Bias in healthcare (list of cases)

Debate: Bias in the Healthcare ecosystem (miro)

First assignment