AI Frontiers: Measuring and mitigating harms with Hanna Wallach

This episode of the Microsoft Research Podcast features a conversation with Partner Research Manager Hanna Wallach, who has extensively researched fairness, accountability, transparency, and ethics in AI and machine learning. Her insights have informed the use of AI in Microsoft products and services for years. Wallach explains how she and her team expanded their tools for measuring fairness-related harms in AI systems to address harmful content more broadly during their work on the Bing Chat deployment. She also discusses her interest in filtering, a technique widely used but not often discussed for mitigating harms. Additionally, cross-company collaboration brings policy, engineering, and research together to execute Microsoft's approach to responsible AI development and deployment.

The episode is part of a series hosted by AI scientist and engineer, Ashley Llorens, who engages in conversations with his colleagues and collaborators about the latest in AI. The series explores what the latest AI models, such as GPT-4, mean for our approach to understanding, creating, and deploying AI. It also examines the potential of AI to benefit humanity, as well as its applications in areas such as healthcare and education. With powerful AI models showing dramatic improvements in reasoning, problem-solving, and language capabilities, AI research is accelerating, and so is its progress.

For more information, check out Microsoft AI: Responsible AI Principles and Approach and AI and Microsoft Research.