Idea: the intelligence explosion convention

Abstract

An immediate governance framework is needed to manage the challenges and risks associated with a rapid technological “intelligence explosion” driven by advanced AI. This framework centers on establishing a global convention triggered by defining a specific, verifiable threshold point. This threshold would be characterized by technical benchmarks and confirmed by a panel of experts, balancing the need to intervene before catastrophic risk and leveraging the availability of advanced AI deliberative assistance. Upon crossing this point, the proposal requires the United States to commit to a one-month pause in frontier AI development, during which it would convene a convention. Nations that verifiably pause development may send delegates to draft multilateral treaties. These treaties would address emergent issues critical during the intelligence explosion, including restrictions on AI development and proliferation, investment in safety, governance of newly unlocked resources (space and terrestrial), protection of democratic structures, securing economic power post-labor, and establishing rights for potential digital entities. This strategy aims to create a necessary political action trigger and capitalize on a brief period before irreversible national power imbalances occur. – AI-generated abstract.