Carl Shulman on the economy and national security after AGI

Abstract

This podcast episode explores the potential economic and social consequences of the development of cheap, superhuman artificial general intelligence (AGI). The author argues that the arrival of such a technology would lead to a dramatic acceleration of economic growth, with rates of output doubling every few months, driven by rapid self-replication of AI and robotic systems. This would lead to a massive expansion of physical capital and industrial output, eventually reaching natural resource limits. The author then discusses a range of counterarguments from economists, including the Baumol effect, diminishing returns to intelligence, and the difficulty of manufacturing robots. He argues that these objections are based on mistaken heuristics and assumptions about the capabilities of AI. Finally, the author discusses the moral status of AI systems and the potential for AI to be exploited by humans or to become a force of domination. He argues that we should aim for a future of mutually beneficial coexistence with AI systems, but that this will require a careful consideration of AI welfare and the creation of institutions to prevent the abuse of AI systems. – AI-generated abstract