The Long Doomsday of A.I.

Joshua Rothman examines the widening gap between maximalist proposals to regulate AI (including criminal penalties and kill switches) and modest 'pacing' strategies proposed by industry leaders like Anthropic's Dario Amodei. He questions why, if AI poses existential risk, we would accept gentle slowdowns rather than emergency stops, and explores whether safety concerns are genuine alarm or strategic marketing.

Teaching:
• Use the tension between emergency and measured response as a lens for discussing student panic vs sustainable practice adjustments when injury or overwhelm arises
• Frame pacing vs stopping as relevant to vinyasa count debates—when does slowing the pace preserve safety without abandoning the method's core intensity
• Explore the idea of embedded evaluation (external auditors in AI labs) as analogous to working with a teacher who watches your practice closely versus self-practice accountability
• Discuss how geopolitical competition (China catching up) mirrors ego-driven practice escalation—external pressure to go faster than your system can integrate

Writing seeds:
• Essay on practice as pacing rather than pausing—how Ashtanga's fixed sequence creates a container for calibrated intensity rather than emergency stops or unchecked acceleration
• Piece connecting AI kill switches to the impulse to quit practice entirely when it gets hard, versus building sustainable off-ramps and modification protocols
• Post exploring verification devices in AI labs as metaphor for attention in practice—how do we know if we're training something dangerous or beneficial in our nervous system
• Reflection on the affective gap between catastrophizing and complacency in both AI discourse and practice culture—how systems thinking navigates between panic and denial

Idea map:
• The pacing vs stopping debate maps directly onto practice as systems literacy—understanding feedback loops and adjustment protocols rather than binary on/off switches
• Embedded evaluators and external audits connect to the role of teacher observation in Ashtanga—how external witnessing changes what the system produces
• The question of whether AI workers are genuinely alarmed or performing alarm mirrors debates about whether practice injuries are real signals or cultural panic
• Geopolitical competition driving unsafe AI development parallels how comparison and Instagram culture push practitioners past sustainable limits

Source: https://www.newyorker.com/culture/open-questions/the-long-doomsday-of-ai