AI Frontiers: Measuring and mitigating harms with Hanna Wallach

This Microsoft Research Podcast episode focuses on the impact of large-scale AI models, specifically GPT-4, and the progress it signals for artificial intelligence as a whole. The host, Ashley Llorens, engages in conversations with his colleagues regarding the implications of these models, the benefits they bring, and their applications in various fields, including healthcare and education. Partner Research Manager, Hanna Wallach, is featured in this episode, discussing her extensive research into fairness, accountability, transparency, and ethics within the realm of AI and machine learning.

Wallach provides insight into the tools she and her team developed to measure fairness-related harms in AI systems, as well as their use in addressing harmful content for the deployment of Bing Chat. She also talks about her interest in filtering, a technique for mitigating harm which is frequently used but often overlooked. The episode highlights the importance of a cross-company collaboration that brings policy, engineering, and research together to develop and deploy AI responsibly, a topic addressed in Microsoft's AI: Responsible AI Principles and Approach.

This episode of the Microsoft Research Podcast offers a valuable glimpse into the future of AI and the role it can play in improving our lives. It also emphasizes the importance of ethical considerations in the development and deployment of AI technologies.