Alyssa emphasizes proactive monitoring for biases in ML models, suggesting setting up real-time data monitoring with human review and alert systems. Lukas and Alyssa discuss improving training data quality by addressing selection bias and ensuring data reflects the intended user community.
