In this study, Dr. Bullock examines a scenario in which the world has created a single AGI that is governed by a specific Global AGI Agency. The report explores the type of AGI that has been created and the institution that governs its use. In this scenario, the Global AGI Agency will be comprised of both the type of AGI and the institution that governs it. Dr. Bullock argues for the development of an Open Agency AGI model, which offers potential benefits in terms of safety, controllability, and transparency compared to Unitary Agent models.
The proposed Global AGI Agency has four core elements: (1) an institutional framework involving a joint support from the UN and IEEE and a coalition of governments and private sector companies, (2) global collaboration and integration across key AI powers and companies, (3) the Open Agency AGI model itself, organized around principles of structured transparency, and (4) mechanisms for democratic accountability and access.
This proposal seeks to leverage existing global governance institutions while incorporating the crucial role of private companies in AI development. While acknowledging the numerous challenges such an agency would face, including international cooperation hurdles and risks of power centralization, this report contributes to the growing literature on international AI governance. It does so by encouraging the creation of a specific form of AGI, assuming a world in which AGI is developed, expanding the stakeholders involved, and proposing specific mechanisms for democratic accountability and access.
The aim is to present an idealized scenario that can inform and guide our approach to the development and governance of this potentially transformative technology.
The final report will be available soon.
