Insider Brief
- Project Canary, a collaboration between RAND and METR, secured approximately $38 million from The Audacious Project to study and mitigate risks associated with advanced AI systems.
- The initiative will develop tools and methodologies to assess the risks of AI systems before deployment, with the goal of making AI systems safer and sharing these resources with the broader AI research community.
- The project, led by RAND and METR researchers, aims to provide insights to policymakers and promote responsible AI development, with a focus on reducing potential misuse and ensuring AI system autonomy.
PRESS RELEASE — Project Canary — a collaboration with METR — is to receive approximately $38 million to study growing risks of Artificial Intelligence
Recognizing the potential impact of advanced artificial intelligence (AI) on society, The Audacious Project, a collaborative funding initiative housed at TED, has committed approximately $38 million to RAND and METR for Project Canary, a new research collaboration designed to help make future AI systems safer.
Selected as one of The Audacious Project’s 2024 cohort of big ideas, the project aims to develop tools that all frontier AI developers can use to evaluate their systems for risks prior to release.
“AI systems are advancing at unprecedented rates. Alongside the incredible potential come significant risks, many of which we don’t yet fully understand,” said RAND President and CEO Jason Matheny. “RAND is researching the implications of AI across nearly every policy domain, but this project is exciting because it is the first to incorporate our evidence-based approach to assess the capabilities of frontier AI systems while still in development.”
Researchers from nonpartisan, nonprofit RAND and METR, an AI evaluation research nonprofit, will each contribute to the project, building on their collective expertise in misuse risk and AI system autonomy, respectively. The project will focus on developing cutting-edge science, methodology, and tools to evaluate how risks from AI systems might manifest, and conducting predeployment evaluations of each new generation of frontier AI systems.
The methodologies and tools created by the project will be shared with the wider AI research community, allowing others to build on existing research and significantly lowering the barrier to entry for newer organizations. RAND and METR will also share their findings and insights with decisionmakers to inform responsible development and deployment of AI systems.
RAND researcher Ella Guest and Beth Barnes, the founder and head of research at METR will lead the research efforts for Project Canary.
See the full scope of RAND’s work on AI here.
About the Audacious Project
Launched in April 2018, The Audacious Project is a collaborative funding initiative that’s catalyzing social impact on a grand scale. Housed at TED, the nonprofit is devoted to ideas, and with support from The Bridgespan Group, The Audacious Project convenes funders and social entrepreneurs with the goal of supporting bold solutions to the world’s most urgent challenges. The funding collective is made up of respected organizations and individuals in philanthropy, including the Bill & Melinda Gates Foundation, ELMA Philanthropies, Emerson Collective, MacKenzie Scott, Skoll Foundation, Valhalla Foundation, and more. Each year The Audacious Project supports a new cohort. The 2024 grantees are Equipo Argentino de Antropolgia, Every Cure, Food4Education, Global Methane Hub, Great Barrier Reef Foundation, Inkomoko, MapBiomas, Project Canary, Scaling Safety, and Transcend Education.
Learn more at AudaciousProject.org.
