Rethinking Algorithmic Fairness
Sharad discusses the limitations of traditional definitions of fairness in algorithms, such as error equal rates and calibration. He emphasizes that merely being calibrated does not equate to equity, using the example of discriminatory lending practices. Advocating for a broader understanding of fairness, he suggests that the focus should shift from mathematical definitions to the real-world consequences of algorithmic decisions.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
The Measure and Mismeasure of Fairness with Sharad Goel - #363
Related Questions