Worried about an AI apocalypse? Here’s how to think about the risk of doom — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- P(doom) estimates from leading AI commentators span 0% to over 95%, covering the full probability spectrum and offering no scientific framework.
- Jensen Huang, Nvidia's CEO, recently said there was a 0% chance of an AI apocalypse by 2030 — a view the author disagrees with, placing the figure at "greater than zero but small."
- Climate change is far more quantifiable: under current policies, a 100% chance of exceeding 1.5°C this century, falling to 79% under the strongest action scenarios, per the UN Environment Programme.
- Asteroid impacts above 1km occur roughly once every 600,000 years; NASA has catalogued over 90% of them, and "city killer" asteroids (140m+) hit every ~20,000 years.
- Pandemic risk sits at a 2-3% chance of a COVID-19-level event every year, based on historical patterns showing major outbreaks every 33-50 years (1918 flu killed ~50 million; HIV killed 40 million-plus).
- Road traffic collisions kill over 1 million people annually — global odds of ~0.01% — though over 90% of deaths occur in low- and middle-income countries, per the WHO.
Why it matters: The piece undercuts the public discourse around AI existential risk by demonstrating that p(doom) figures come with no scientific basis, while quantifiable threats — climate change (100% chance of exceeding 1.5°C under current policies) and pandemics (2-3% chance yearly) — deserve more policy attention than AI apocalypse scenarios that experts estimate anywhere between 0% and 95%.
Ask SkimNews
