Gregory C. Allen and Georgia Adamson
1. OVERVIEW
On November 21 and 22, 2024, technical artificial intelligence (AI) experts from nine countries and the European Union will meet for the first time in San Francisco. The agenda: starting the next phase of international cooperation on AI safety science through a network of AI safety institutes (AISIs). The United States, United Kingdom, European Union, Japan, Singapore, South Korea, Canada, France, Kenya, and Australia make up the initial members of the network, which was first launched by U.S. secretary of commerce Gina Raimondo at the May 2024 AI Seoul Summit. At the time of the launch, Italy and Germany were also potential members of the network, as signatories to the Seoul Statement of Intent toward International Cooperation on AI Safety Science, or Seoul Statement, the network’s founding document. However, a September announcement by Raimondo and U.S. secretary of state Antony Blinken confirmed that Kenya would instead be the final member of the AISI International Network at this stage.
According to the Seoul Statement, the international network will serve to “accelerate the advancement of the science of AI safety” at a global level by promoting “complementarity and interoperability” between institutes and fostering a “common international understanding” of AI safety approaches. While the statement does not define specific goals or mechanisms for AISI collaboration, it suggests that they “may include” coordinating research, sharing resources and relevant information, developing best practices, and exchanging or codeveloping AI model evaluations. Now, in the months following the AI Seoul Summit, AISI network members must begin to articulate the objectives, deliverables, timelines, and avenues for cooperation that will put the promise of AISI cooperation into action.
No comments:
Post a Comment