Welcome to the second episode of the ChinAI podcast, hosted by Jeff Ding. Our guest today is Eric Drexler, a senior research fellow at the Future of Humanity Institute. He joins the ChinAI podcast to discuss the latest FHI technical report “Reframing Superintelligence: Comprehensive AI Services as General Intelligence.” In contrast to conventional views of superintelligence as an agent with unbounded capabilities across multiple domains, Eric reframes superintelligence as embodied within a bounded framework of Comprehensive AI Services (CAIS). We discuss how this alternative model may lessen some of the classic risks associated with artificial general intelligence (e.g. the paperclip maximizer) but also bring under-explored risks to the fore (e.g. supercharged addiction).
Often described as the founding father of nanotechnology, he provides a unique perspective on the pathway to superintelligence — one of an extremely perceptive systems engineer. We also examine a perspective on superintelligence from a prominent Chinese philosopher.
*****Timestamps: Briefing Checklist (1:00); Debate the Guest (20:00); Footnote Fever (31:00); Trust the Process (37:50)