Jesuit Perspectives: Algorithm Biases
Algorithmic bias is a term that describes the fact that computer programs that use machine learning can be discriminatory, or unfair. Algorithms are step-by-step processes that are written in computer programming code. Machine learning is a term that refers to computer programs that identify patterns and relationships within data, learning these patterns and relationships from the data itself. A common application of machine learning is to classify observations, that is to determine if observations belong to a specific category or not. Since machine learning algorithms learn to classify by being trained on data, if the data in the data sets that are used to train the algorithms are not representative of the population, then the classifications made by the algorithms can be biased. Of particular concern is that machine learning algorithms have been shown to be biased against people of different races and genders.
This document provides a set of teaching resources, including questions, discussion points and instructor notes for incorporating Jesuit perspectives and values in the classroom. These resources are intended to be used as supplementary materials for Information Systems/Analytics courses. The resources are instructor-focused and meant for the instructor and not the students. This series from Inspirational Paradigm presents teaching resources focused in information systems and analytics with five categories including: Data for Good (DFG), Income Inequality (INI), Algorithm Biases (AB), Privacy (P), and Ethics (ETH). The document available on this page focuses on Algorithm Biases (AB).
- Recognize the social justice issues related to IT and data analytics: digital divide, data and algorithm biases, and income inequality.
- Understand the technology ethical issues (loss of privacy, rumors & misinformation; data ownership and data access)
- Understand the role IT and IS play for societal good (ability to inform public about health, finance, services, crisis management, environmental sustainability, etc.)