Panel: Gender bias in AI - In collaboration with Women in Data Science Belgrade
summary
Panelists describe concrete interventions and failures: a racing game removed gender field so analysis of male vs female behavior disappears; a lending model still discriminates because zip code as proxy for race; teams pilot fairness metrics, counterfactual tests and synthetic data to rebalance training sets; regulators and GDPR‑style rules are proposed as deployment gates. Recommendations are specific: improve data quality, add transparency and observability, prioritize female hiring and leadership, iterate human-in-the-loop alignment and accept tradeoffs between accuracy, fairness and business speed.
AI-written summary · about
chapters (10)
- 0:00Panel introduction
- 1:12About WiDS Belgrade
- 2:37Algorithmic bias in AI
- 8:47Data quality and bias
- 13:05Diversity in AI teams
- 19:06Frameworks for bias
- 25:04Biased AI and inequalities
- 31:27Mitigating gender bias
- 37:12Future of gender in AI
- 41:52Q&A on bias and ethics
related talks
44mPanel: AI Feminism
01h 27mPanel - Women in Data Science
43mDiversity: Two steps forward, one step back (just keep it going)
47mHow gender imbalance is holding back the future of AI
34mAI and Gender: Building Fair AI Systems for the Future
54mHow diversity drives excellence in our data driven tech world
27mCrouching Harm, Hidden Bias—Addressing issues of fairness....
38mSocial Implications of Bias in Machine Learning