The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Privacy vs Fairness in Computer Vision with Alice Xiang - #637
8 snips
Jul 10, 2023 Alice Xiang, a Lead Research Scientist at Sony AI and Global Head of AI Ethics at Sony Group Corporation, shares her expertise on the critical issues of privacy and fairness in computer vision. She discusses the impact of data privacy laws and the dangers of unauthorized data use, emphasizing the importance of ethical practices in AI. Alice highlights the history of unethical data collection and the challenges posed by generative technologies. Solutions such as community engagement and interdisciplinary collaboration are also explored, alongside the need for robust AI regulation.
AI Snips
Chapters
Transcript
Episode notes
Early AI Ethics Experience
- Alice Xiang's first commercial machine learning model highlighted potential data skews and bias.
- This experience motivated her to explore AI ethics from technical, legal, and policy perspectives.
Affirmative Action and Algorithmic Fairness
- Algorithmic fairness and affirmative action policies intersect in their approaches to bias mitigation.
- Legal debates around affirmative action inform discussions on algorithmic fairness and what constitutes fairness.
Seen vs. Unseen
- There's tension between privacy and fairness in computer vision, particularly regarding data representation.
- Underrepresented groups may not want increased representation in datasets, creating a 'seen versus misseen' dilemma.