Responsible AI Engineering

Catalog description

The deployment of modern AI systems presents novel challenges, including bias amplification, ensuring safe autonomy, addressing hallucinations, and more. This course teaches engineering AI-enabled software systems that adhere to key requirements of responsible AI: fairness, robustness, explainability, and safety. In addition to optimizing for accuracy, future software engineers and data scientists would need to develop tools and methodologies for testing, verifying, and maintaining machine learning components to meet these principles. The course thus explores evaluation and mitigation strategies designed to ensure the responsible use of ML models, deep neural networks, and generative AI in safety-critical applications. Students will engage with recent research in responsible AI, gaining insights into the latest advancements and challenges in the field. The course also includes a hands-on project, enabling students to apply their knowledge to a real-world problem that aligns with responsible AI. Offered as CSDS 347 and CSDS 447 . Prereq: Graduate Student standing or CSDS 440 or CSDS 340 or CSDS 391.