Stanford Seminar - Responsible AI (h)as a Learning and Design Problem

December 6, 2024 Michael Madaio, Google Research To address the potential harms of AI systems, prior work has developed resources (e.g., toolkits) to support responsible AI (RAI) development and studied how AI practitioners use such resources in their development practices. However, recent work suggests that AI practitioners may not have the relevant skills or knowledge to effectively use RAI resources--particularly as pre-trained AI models have made it easier for more people to build AI-based applications. In this talk, I will share findings from my recent research on 1) what and how AI practitioners in industry contexts are learning about responsible AI on-the-job, and 2) opportunities to support practitioners' in situ learning during AI design. I will close with implications of our findings and open questions for how HCI, design, and the learning sciences might contribute to the responsible design and development of AI. About the speaker: Michael Madaio is a Senior Research Scientist at Google Research. His current research draws on methods from human-computer interaction to help AI practitioners responsibly design AI systems. Prior to joining Google, he was a postdoc at Microsoft Research’s FATE research group on fairness, accountability, transparency, and ethics in AI, and he completed his Ph.D. in Human-Computer Interaction from Carnegie Mellon University, where he was a fellow in the Institute for Education Sciences' Program for Interdisciplinary Education Research (PIER). His research has received several best paper awards, including at the ACM Conference on Fairness, Accountability, and Transparency (FAccT), the ACM Conference on Human Factors in Computing Systems (CHI), the International Conference of the Learning Sciences (ICLS), and others. More about the course can be found here: https://hci.stanford.edu/seminar/ View the entire CS547 Stanford Human-Computer Interaction Seminar playlist:    • Stanford CS547 - Human-Computer Interactio...   ► Check out the entire catalog of courses and programs available through Stanford Online: https://online.stanford.edu/explore