Microsoft Research Podcast Ideas: Community building, machine learning, and the future of AI
11 snips
Dec 1, 2025 
Hanna Wallach, a leading researcher in computational social science and responsible AI, shares insights from her experience co-founding the Women in Machine Learning workshop. She discusses the evolution of WiML over 20 years, highlighting community challenges and successes. The conversation dives into the gaps between theoretical fairness in AI and real-world applications, and how generative AI demands rigorous evaluation. They also emphasize the importance of designing AI that fosters critical thinking and shares valuable advice for aspiring researchers.
AI Snips
Chapters
Transcript
Episode notes
Small Peer Rituals Sustained Early Careers
- Both speakers recall building small supportive communities like 'ladies' brunch' during male-dominated PhD days.
- These informal networks helped sustain them through lonely training environments.
Practitioners Didn’t Fit Academic Fairness Assumptions
- Their FairLearn practitioner study revealed practitioners used many non-predictive ML approaches and faced data-collection barriers.
- This led Jenn Wortman Vaughan and colleagues to publish a paper characterizing the research-practice gap in fairness tools.
How WIML Began From A Hotel-Room List
- Hanna Wallach and peers found only about 10 women in ML at NeurIPS and organized their own event after a Grace Hopper rejection.
- Their bold student-run workshop drew ~100 women and became a full day of research talks.