Which end of the world should you actually worry about?
← All theories

Engineered or natural pandemic

Rank #1 Health

A new pathogen spreads worldwide and kills at least 10 million people within about two years, far beyond recent outbreaks.

Round two, with fewer sourdough starters.

Combined score

24.3

Unlikely

Average of 4 AI models, as the chance this happens within 25 years. Soonest window most models chose: 1-5 years.

Last updated .

Independent AI panel

Claude (Anthropic)

12

Unlikely | soonest: 1-5 years

  1. Only about one clear historical case (1918 influenza) fits 10M+ deaths in roughly two years. COVID-19 is borderline, with official deaths near 7M.
  2. A 25-year window with a rough 1-in-100-years base rate gives a probability in the 10-25% range, before adjusting for the strict two-year, new-pathogen requirement.
  3. Wikipedia attention to 'Pandemic' and 'Biological warfare' is flat to slightly down. It signals public interest, not a change in underlying risk. GDELT data was unavailable.

GPT (OpenAI)

15

Unlikely | soonest: within 1 year

  1. Past pandemics show that worldwide spread and deaths on this scale are possible, though rare.
  2. Modern surveillance and medical countermeasures may limit spread and mortality, but cannot eliminate the risk.
  3. The digest shows no clear warning signal; page-view counts offer little evidence about pandemic likelihood.

Gemini (Google)

35

Possible | soonest: 1-5 years

  1. Advances in synthetic biology and low-cost lab equipment increase the risk of accidental or deliberate release of a high-consequence pathogen.
  2. Historical patterns show pandemics like the 1918 flu occur on a scale that aligns with the 10 million+ death threshold within a 25-year window.
  3. Increased global travel and urbanization significantly facilitate the rapid, worldwide transmission required to meet the severity of this theory.

Grok (xAI)

35

Possible | soonest: 5-25 years

  1. Wikipedia page views for pandemic topics are moderate and declining, showing no surge in public concern.
  2. Modern surveillance and vaccines lower the odds of a 10M+ death event compared to historical precedents.
  3. Data digest lacks signals on lab risks or outbreaks, supporting only a modest 25-year probability.

Signals

Wikipedia page views: Pandemic
Wikimedia pageviews API, as of 2026-11-16
326/day over the last 7 days (30 day average 343/day, -7% vs the earlier weeks)
Wikipedia page views: Biological warfare
Wikimedia pageviews API, as of 2026-11-16
308/day over the last 7 days (30 day average 321/day, -5% vs the earlier weeks)

Questions people ask

Could a new pandemic kill 10 million people?
The combined score is 24.3 out of 100 (Unlikely), the average of 4 AI models. Each model estimates the chance of this exact event happening within the next 25 years. None declined.
How soon could a pandemic that big happen?
The soonest window most models chose is "1-5 years" (2 of 4 scoring models). That is when they think it could plausibly happen at the earliest, not when it will.
What do AI models say about pandemic risk?
Claude said: "Only about one clear historical case (1918 influenza) fits 10M+ deaths in roughly two years. COVID-19 is borderline, with official deaths near 7M." GPT said: "Past pandemics show that worldwide spread and deaths on this scale are possible, though rare." Gemini said: "Advances in synthetic biology and low-cost lab equipment increase the risk of accidental or deliberate release of a high-consequence pathogen." Grok said: "Wikipedia page views for pandemic topics are moderate and declining, showing no surge in public concern." Entertainment plus real data. Not a prediction service.

How it works

Every model gets the same data summary and rates this exact event over the next 25 years. The number maps to a band, and the ranking uses the average. Read the full method.

Entertainment plus real data. Not a prediction service.