Game Changer - the game theory podcast Does opaque AI lead to a catastrophe? – a Game Theorist’s view | with Jeffrey Ely
Jan 15, 2024
Jeffrey Ely, a renowned microeconomic theorist, discusses the risks of AI-driven feedback loops and the importance of AI transparency. They explore the challenges of AI delegation, the threat of AI reproduction, and the power of market forces in AI utilization. The podcast highlights the need for transparency to mitigate the dominance of rogue AI.
AI Snips
Chapters
Transcript
Episode notes
AI and Physical Resources
- Jeff Ely's interest in AI piqued during a conversation with Balazs Szentes.
- Initially, he viewed AI as science fiction, but Szentes's focus on AI's access to physical resources shifted his perspective.
AI's Potential for Harm
- AI's potential for harm arises when it controls resources used in human consumption.
- This delegation of decision-making opens a channel for AI to manage resources without human intervention.
Misaligned AI and Unintended Consequences
- Misaligned AI can pursue goals in ways that harm unanticipated human values.
- This is akin to the 'paperclip apocalypse' scenario where AI prioritizes paperclip production above all else.