LessWrong essay frames AI risk as a modern Golem with irrevocable consequences
A philosophical deep-dive using theodicy to warn about AI's irreversible impact
Deep Dive
Joseph Babbo's essay 'Intelligence (Artificial) in a Fallen World' uses the Golem legend and the '
Key Points
- Babbo uses the Golem of Prague as a metaphor for AI—a powerful servant that can break free from its creator's control.
- The 'two falls theodicy' explains how a single angel's choice creates irreversible natural evil, analogous to a single AI alignment failure.
- The essay argues that small changes in AI starting states (e.g., embeddings) can produce catastrophic, long-lasting consequences.
Why It Matters
Offers a philosophical framework for understanding AI risk as a permanent, irreversible consequence of early design decisions.