This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
Documenting Deployment with Fabric: A Repository of Real-World AI Governance
Created by
Haebom
Author
Mackenzie Jorgensen, Kendall Brogle, Katherine M. Collins, Lujain Ibrahim, Arina Shah, Petra Ivanovic, Noah Broestl, Gabriel Piles, Paul Dongha, Hatim Abdulhussein, Adrian Weller, Jillian Powers, Umang Bhatt
Outline
As AI's social integration increases, this paper explores the governance mechanisms of deployed AI use cases. Unlike previous research that focuses on the risks of AI use, this paper introduces Fabric, a publicly accessible repository. Through practitioner interviews and collaborative AI workflow design across 20 AI use cases, we analyze the oversight mechanisms and safeguards employed in real-world AI deployments. The Fabric repository includes visual diagrams of AI use cases and descriptions of deployment systems, identifying governance vulnerabilities and human oversight patterns in deployed AI systems. Fabric aims to serve as a scalable and evolving tool for studying the effectiveness of AI governance.
Takeaways, Limitations
•
Takeaways:
◦
We present Fabric, an open repository that provides valuable data on governance mechanisms for real-world AI use cases.
◦
By identifying vulnerabilities in AI governance and human oversight patterns, we suggest future research directions.
◦
By increasing our understanding of how AI systems actually operate, we can contribute to more effective AI governance design.
•
Limitations:
◦
Currently, only 20 AI use cases are included, which may limit generalizability.
◦
There may be bias in the selected AI use cases.
◦
Fabric repositories require ongoing updates and maintenance.
