Chat, are we doomed?
Sep 24, 2026 · 1h 13m
More Science podcasts · More News podcasts
Summary
Jerusalem Dempis and Matthew Glia debate the existential risks of advanced AI, specifically the threat of recursive self-improvement leading to human extinction. Dempis argues that as AI becomes flawlessly capable, humans will cede control to it, creating an unrecoverable scenario, while Glia remains skeptical, citing the gap between digital intelligence and physical robotics.
Topics discussed
Sponsors: BKFC 94 and BKFC Sea Cruise
Intro: Matt's concern about AI extinction risk
Comparing AI risk to asteroid and extinction events
Skepticism on recursive self-improvement and human oversight
Gradual automation of decision-making in military and daily life
The paradox of AI safety: better AI reduces accidental harm
Public attention and the need for a 'Hiroshima' moment
Sponsor: Forkful meal delivery service
Sponsors: BKFC 94 and BKFC Sea Cruise
Why focusing on proximate risks may not solve existential threats
The 'parenting' analogy for managing superintelligent AI
Political dynamics and the role of international treaties
Critique of AI doom discourse and media framing
Historical context: George Eliot and the self-improvement hypothesis
Skepticism on RSI, superintelligence, and robotics bottlenecks
The 'Chimpanzee' problem: Can humans conceive of superintelligence?
Comparing AI risk to nuclear winter and biological weapons
The 'Machine God' theology and secular humanist concerns
Clarifying nuclear winter and the Cold War context
AI pauses, antitrust exemptions, and Chinese alignment
Political cooperation and the limits of frontier tech optimism
Sponsor: BKFC Sea Cruise
Study: Why meetings are actually beneficial for work
More from The Argument
Listen ad-free on Castria