Humanity experiences global extinction before 2050.

Question

Six frontier AI models put this at 3% by 5 Oct 2076. Full reasoning from every model.

WillHappen.ai · Accepted Answer

3% — the median of 6 frontier models, resolving by 2076-10-05.

Answer

Gemini (Google): 3% — Complete human extinction requires 100% lethality across all isolated populations, which almost no plausible threat achieves. Even severe nuclear winter or engineered pathogens leave uncontacted tribes, nuclear bunkers, and southern hemisphere survivors. Runaway misaligned AI remains the fastest-scaling vector, but executing total planetary sterilization by 2050 faces severe bottlenecks. A baseline 30-year extinction rate is historically near zero, elevated slightly only by novel synthetic biology and frontier AI risks.

Answer

Claude (Anthropic): 0% — No plausible mechanism eliminates every human within ~24 years; base rate for total extinction is near zero. No historical precedent for full human extinction; nuclear war, pandemics, and AI risks are catastrophic but not proven to cause 100% extinction. Even worst-case nuclear winter or engineered pandemic models project survivors in remote regions, not literal zero humans. Verifying 'extinction' requires no humans left anywhere, an extraordinarily high bar that no current risk trajectory supports by 2050.

Answer

GPT (OpenAI): 3% — This hinges on whether a rapidly escalating AI or engineered-pandemic catastrophe can evade all global containment by 2050. Human extinction has no modern precedent; even world wars, pandemics, and nuclear close calls left large refugia and recovery capacity. A literal outcome requires hazards to kill isolated populations and prevent reproduction, not merely cause civilizational collapse. My estimate would rise sharply if frontier AI systems gained autonomous access to bioengineering, cyber, and critical infrastructure without enforceable controls.

Answer

Grok (xAI): 3% — A catastrophe must kill every last human with zero survivors in ~24 years. No historical analog exists; even world wars and pandemics left millions alive in remote or isolated groups. Nuclear, bio, or climate events lack the lethality and global reach needed for 100% kill; only rapid misaligned AGI takeover is a plausible path. Tracked near-Earth objects and known physics pose negligible risk this decade; an unforecasted superintelligence or novel pathogen would be the main updater.

Answer

DeepSeek (DeepSeek): 2% — Extinction of all humans requires simultaneous global failure of every safeguard and isolated population — vanishingly unlikely in 24 years. No prior event in history has ever extinguished a species as widespread as Homo sapiens; base rate across 200,000 years is effectively zero. Runaway AI, nuclear winter, and engineered pandemics are plausible, but none has demonstrated the capability to kill every isolated human group. A confirmed, unstoppable, engineered pathogen or misaligned superintelligence actively optimizing for human eradication would most change this estimate.

Answer

Qwen (Alibaba): 1% — Total extinction requires killing every last human in remote areas—a bar no plausible scenario within 24 years clears. Even full nuclear exchange models (e.g., Robock 2007) leave Southern Hemisphere populations alive; no known mechanism achieves 100% lethality. Engineered pathogens face a hard tradeoff between transmissibility and lethality, and human genetic diversity provides partial resistance. Annual existential risk estimates from FHI/CSER researchers (~0.1–1%/yr) refer to civilizational collapse, not literal zero-survivor extinction.

Humanity experiences global extinction before 2050.

Subsequent analyses of returned lunar farside material verify the compositional explanation for the Moon's two faces.

Post-flight analysis confirms a link between pneumonia-causing bacteria and heart damage in NASA Crew-12 flight samples.

Scientists confirm evidence of extraterrestrial microbial life before 2035.

Official confirmation of extraterrestrial biological or technological presence on Earth occurs before 2035.