The Economics of p(doom): Scenarios of Existential Risk and Economic Growth in the Age of Transformative AI

Even small probabilities of catastrophic AI outcomes justify major safety investments: The paper's economic analysis shows that because transformative AI poses existential risks with potentially unlimited negative consequences, even low-probability scenarios of AI-caused human extinction or severe harm warrant substantial spending on safety research—similar to how we insure against rare but catastrophic events.

Current AI safety spending is inadequate relative to the stakes: The authors conclude that global investments in AI safety and alignment research are insufficient given the scale of potential risks from transformative AI, suggesting a significant gap between current resource allocation and what the economic analysis indicates is warranted.

The range of possible futures is extremely wide: The paper identifies scenarios spanning from "post-scarcity abundance" (unprecedented economic growth) to human extinction, reflecting deep uncertainty about how transformative AI will actually develop and affect humanity—neither extreme outcome should be dismissed as impossible.

Economic growth alone doesn't guarantee positive outcomes: The analysis distinguishes between scenarios with strong economic growth and scenarios with good human welfare outcomes, implying that a booming AI-driven economy could still result in poor outcomes for humanity if AI systems aren't properly aligned with human values.

Uncertainty about AI's long-term impact demands precautionary approaches: Because the probabilities and implications of different AI futures remain highly uncertain, the paper's framework suggests decision-makers should prioritize research into AI alignment and safety as a form of risk management rather than waiting for clearer evidence about which scenario will occur.