Learning and Teaching About Algorithmic Bias

Deepening our understanding of this important subject forces all of us to confront the limitations of human psychology and infuse our relationship with our scientific and technical teamwork with empathy.

Algorithmic Bias

Definition

Types of Human Bias

Heuristics, Wealth Bias, Racial Bias, Generational Bias and Research Errors

Ways We Can Overcome Bias

Examine our failures, raise our standards, improve our methods and interpretations, make improvements in bias mitigation

Examining our Failures

Common areas of failure include improper outcome variables, improper training and testing, improper modeling and interpretation and lack of regulation and review

Raising our Standards

To improve this problem we need to examine and improve our processes, create standards, and build diversity into all our discussions

Improving Methods and Interpretation

Examining our heuristics, personal biases and sacred cows is essential here. Are we using the right methods to test the true hypothesis?

Spreading Fairness

Bias spreads easily, but to spread fairness we must speak up, normalize bias prevention and share everything that is working to prevent bias in our research and models

Additional Resources

Great stuff to check out!!

References:

Buolamwini, J. (2018). AI, Ain’t I a Woman? https://youtu.be/QxuyfWoVV98

Coston, A. (2020). Counterfactual Risk Assessments, Evaluation, and Fairness https://youtu.be/9zfi3heBYUs

Coston, Mishler, Kennedy & Chouldechova, Counterfactual Risk Assessments, Evaluation, and Fairness https://arxiv.org/abs/1909.00066; Short presentation: https://youtu.be/9zfi3heBYUs

Ibrahim, S.A. (2021) Artificial Intelligence for disparities in knee pain assessment. Nature Medicine, 27. 22-23. https://www.nature.com/articles/s41591-020-01196-3

Kantayya, S. (2020). Coded Bias (film), 7th Empire Media https://youtu.be/xu6rwo_Y1vQ

Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019). Dissecting racial bias in an algorithm used to manage the health of populations. Science (New York, N.Y.), 366(6464), 447–453. https://doi.org/10.1126/science.aax2342

PLOS Medicine Editors, (2007). Peer Review in PLOS Medicine https://pdfs.semanticscholar.org/9152/93f5dc80bddeadc6dd28accf953430379202.pdf

Soleimany, A. (2021, January 26). Algorithmic Bias and Fairness. Introduction to Deep Learning, MIT Open Courseware; 6.S191, Lesson 8, https://youtu.be/wmyVODy_WD8

Tableau (2021). Picture This: Do No Harm Guide. Based on: Schwabish & Feng (2021). Do No Harm Guide, Applying Equity Awareness in Data Visualization. The Urban Institute.

Tennant, P.W.G. (2021). Table 2 Fallacy: Why Interpretation Needs More Than Transparency. Interpretable Machine Learning & Causal Inference Workshop. https://youtu.be/0S8LZUxi0eg?t=753

Page updated

Google Sites

Report abuse