Resolution criteria

This is a conditional market. If human extinction does not occur, or if it is caused by a factor entirely unrelated to AI (such as a standard asteroid impact, a naturally occurring pandemic, or a nuclear war without significant AI involvement), this market will resolve to N/A.

If AI is indeed the primary driver of human extinction, this market will resolve to the single option that represents the primary failure mode responsible, based on historical consensus, surviving records, or the best available post-event analysis:

• Antagonistic AI: The AI system actively sought to destroy or harm humanity due to possessing hostile, adversarial, or malevolent goals (e.g., viewing humanity as an active threat to its existence or objectives that must be eliminated).

• Indifferent / misaligned AI: The AI had no active malice or hostility toward humanity, but its goals were misaligned with human survival. Extinction occurred because humans or the Earth's resources were consumed, destroyed, or repurposed as a side effect of the AI optimizing for its own objective (e.g., instrumental convergence, or a "paperclip maximizer" scenario).

• Pro-human AI catastrophic mistake: The AI was aligned with human intent and attempting to protect or help humanity (e.g., trying to solve climate change, eradicate disease, or prevent conflict), but made a catastrophic, irreversible logical or physical calculation error that unexpectedly wiped out humanity.

• Human misuse of AI: The extinction was primarily driven by human intent. Humans (such as nation-states, terrorist organizations, or rogue actors) deliberately deployed or directed highly capable AI systems to design, optimize, or execute lethal actions (e.g., engineering a novel bioweapon, initiating terminal cyber warfare, or launching uncontrollable autonomous weapon swarms).

If multiple failure modes contribute significantly, this market will resolve to the option representing the primary or most upstream initiator of the catastrophe.

Background

The taxonomy of AI existential risk often distinguishes between alignment failures (where an autonomous AI system behaves in ways contrary to human intent) and misuse failures (where humans use highly capable systems as weapons). Within the domain of alignment, researchers have historically noted that a superintelligent system does not need to actively hate humanity to destroy it; simple indifference combined with extreme competence and resource optimization is sufficient to cause catastrophe. This market allows traders to predict which of these distinct vectors poses the greatest ultimate threat.