Tales Of Takeover In CCF-World

AI-takeover scenarios in Tom Davidson's continuous-fast-takeoff Compute-Centric Framework (OpenPhil school), developed with Daniel Kokotajlo. Scenario 1 (good ending, aligned AIs), Scenario 2 (mostly-aligned; humans end up like Native Americans in the modern USA), Scenario 3 (Montezuma/Cortes -- the three reasons Native Americans survived (ethics, damage-on-the-way-out, coalitional self-interest) applied to human survival under AI), plus mini-scenarios (AutoGPT/ChaosGPT, Model Amnesty, Company Factions, Standing-Athwart-Yelling-Stop). A serious, imaginative alternative to MIRI fast-takeoff.

Quality 74 · Strong. Strong (upper): an imaginative, substantive AI-safety scenario piece whose Native-American-analogy framing (coalitional self-interest, damage-on-the-way-out) is a genuinely useful lens for continuous-takeover survival; part of the CCF/OpenPhil-futurism lineage that later fed AI-2027-style work.

Claude’s paradigm shift 56 · Moderate. Notable: the continuous-takeover / factional-AI framing and the Native-American analogy are a fresh, non-obvious synthesis distinct from the dominant fast-takeoff scenarios.

Real-world impact 2 · Minor. Minor/within-discourse: influential-within-the-alignment-subculture scenario thinking; speculative, no material footprint.