AI with a Conscience: Building Fair and Transparent Systems in Python

Explore the challenges of fairness in AI systems in this 29-minute PyCon US talk that addresses how Python can help build more equitable machine learning models. Learn how historical inequities, imbalanced labels, and skewed representations in data can lead to unfair outcomes in AI applications. Discover practical methods to measure fairness using metrics like demographic parity, equalized odds, and equal opportunity through the fairlearn library. Through real-world examples, understand how to mitigate biases using techniques such as GridSearch and ThresholdOptimizer. Suitable for Python users of all experience levels, from beginners to experts in machine learning, this presentation provides valuable insights into creating AI systems that treat all individuals and groups equitably.