Wikimedia Research/Showcase/Archive/2021/06
June 2021
[edit]- Theme
- AI model governance
June 23, 2021 Video: YouTube
- Bridging AI and HCI. Incorporating Human Values into the Development of AI Technologies
- By Haiyi Zhu (Carnegie Mellon University)
- The increasing accuracy and falling costs of AI have stimulated the increased use of AI technologies in mainstream user-facing applications and services. However, there is a disconnect between mathematically rigorous AI approaches and the human stakeholdersâ needs, motivations, and values, as well as organizational and institutional realities, contexts, and constraints; this disconnect is likely to undermine practical initiatives and may sometimes lead to negative societal impacts. In this presentation, I will discuss my research on incorporating human stakeholdersâ values and feedback into the creation process of AI technologies. I will describe a series of projects in the context of the Wikipedia community to illustrate my approach. I hope this presentation will contribute to the rich ongoing conversation concerning bridging HCI and AI and using HCI methods to address AI challenges.
- * Slides on figshare
- By Haiyi Zhu (Carnegie Mellon University)
- ML Governance. First Steps
- By Andy Craze (Wikimedia Foundation, Machine Learning Team)
- The WMF Machine Learning team is upgrading the Foundation's infrastructure to support the modern machine learning ecosystem. As part of this work, the team seeks to understand its ethical and legal responsibilities for developing and hosting predictive models within a global context. Drawing from previous WMF research related to ethical & human-centered machine learning, the team wishes to begin a series of conversations to discuss how we can deploy responsible systems that are inclusive to newcomers and non-experts, while upholding our commitment to free and open knowledge.
- * Slides on figshare
- By Andy Craze (Wikimedia Foundation, Machine Learning Team)