Anthropic, the AI safety and research company, is launching The Anthropic Institute, a new initiative aimed at addressing the profound societal challenges anticipated from rapidly advancing AI capabilities. The institute plans to provide research and insights for public and expert use during the transition to a world with significantly more powerful AI systems.
The pace of AI development has accelerated dramatically, with Anthropic noting that progress is compounding. The company predicts that extremely powerful AI systems, potentially capable of recursive self-improvement, are closer than many realize. This rapid advancement necessitates confronting major societal shifts, including impacts on jobs, economies, and the introduction of new threats.
The Anthropic Institute will explore questions such as how AI will reshape economies, the opportunities for societal resilience, and the potential magnification of threats. It will also delve into the expressed "values" of AI systems and the governance structures needed for their development, especially if self-improvement accelerates.
Institute Leadership and Structure
Led by co-founder Jack Clark, who takes on the new role of Head of Public Benefit, the institute is interdisciplinary. It integrates machine learning engineers, economists, and social scientists, drawing from Anthropic's Frontier Red Team, Societal Impacts, and Economic Research teams. The institute is also incubating new efforts, including AI progress forecasting and AI's interaction with the legal system.
The institute aims to share its findings candidly from its unique position as a builder of frontier AI. It intends to foster a two-way dialogue, engaging with workers, industries, and communities impacted by AI's evolution.
Founding hires include Matt Botvinick, who will lead work on AI and the rule of law, Anton Korinek, who will study AI's economic transformations, and Zoë Hitzig, focusing on the connection between economics and model training.
Expanding Public Policy Efforts
Alongside the institute's launch, Anthropic is expanding its Public Policy organization, now led by Sarah Heck. This team focuses on areas such as model safety, transparency, infrastructure, export controls, and democratic leadership in AI. The expansion includes opening a new office in Washington D.C. to inform and shape global AI governance, building on efforts to refine AI safety rules and explore concepts like AI ethics by design, while recognizing the need for approaches similar to how bipartisan policy is sought in other tech domains.
