Mapping and Understanding the Landscape of Tools for AI Governance

[Status: Ongoing]

This project aims to map the growing landscape of tools that support the implementation of ethical AI and develop a deeper understanding of the factors impacting their development, use, and effectiveness. Through a multi-phased methodology including a meta-review, systematic reviews, and stakeholder consultations, this project will explore the breadth of tools available, develop frameworks for evaluation and validation of tools, and deepen understanding of their role in regulatory schemes and the broader AI governance ecosystem. The outcomes of this project will include actionable resources for diverse stakeholders.

Project Team
Effy Vayena
Sara Kijewski
Claire McBride
Heidi Lee

Project Collaborators and Partners
Elettra Ronchi

Description
This project’s focus is on practical tools for implementing AI governance. There is currently a plethora of tools that claim to support ethical AI governance. Some tools are developed by or in partnership with governments, international organizations, or big tech companies, while others have been created by smaller non-profit initiatives or startups. Despite the impressive number of tools available, and their often ambitious aims, we still lack frameworks to evaluate such tools, measure their effectiveness and ultimately demonstrate their contribution to AI governance. Furthermore, the landscape of tools can be hard to navigate due to lack of standardized terminology for approaches, methods, or aims and roadblocks to adoption including lack of awareness, skills gaps, and limited incentives for use. In this project we want to address these unanswered questions by mapping the landscape of tools, empirically investigating their present development and use, and establishing a framework for evaluating AI governance tools and informing their future use and development.

Kijewski, S., Ronchi, E. & Vayena, E. The rise of checkbox AI ethics: a review. AI Ethics 5, 1931–1940 (2025). external page https://doi.org/10.1007/s43681-024-00563-x

To come.