AI Risk
Extinction, Severe Dystopia, and Extreme Catastrophe
Intelligence is not a guarantee of benevolence. This paper asks what happens when civilization-relevant superintelligence arrives, distinguishing extinction, enduring domination, and catastrophe from which recovery remains possible.
The argument
The analysis follows motivation, strategic opportunity, defensive responses, and successor systems. Aligning one powerful system is different from maintaining protection across a civilization of increasingly capable systems. Even a protective regime can become a severe dystopia if it eliminates meaningful human agency.
How to read this assessment
The roughly 52% assessment of extinction or enduring severe dystopia is conditional on the specified rapid transition to superintelligence. It is a subjective model output, not an unconditional forecast or a measured frequency.
A question worth pursuing
What would make protection durable as more powerful successor systems appear?
Cite and share
This page provides an editorial introduction. Consult the paper for its complete evidence, assumptions, sources, and collaboration disclosures. Downloads preserve the edition retrieved on October 18, 2026.