Existential risk

The
Precipice.

A diagram of the existential risks humanity faces this century, after Toby Ord's The Precipice: Existential Risk and the Future of Humanity (2020). His estimate: roughly 1 in 6 chance of an existential catastrophe by 2120 — with most of the danger from technologies of our own making.

Source
Toby Ord, The Precipice
Horizon
Next 100 years
Total risk
~1 in 6 (~17%)

Humanity at the edge: the choices of this century determine whether there is a 30th.


The headline number

A 1 in 6 chance
this century.

Ord's central estimate of the probability of an existential catastrophe — humanity going extinct, civilization collapsing permanently, or our long-term potential being foreclosed — in the next 100 years.

For comparison: rolling a die. One face, of six, ends the human story. He calls it a conservative estimate.

×
Five futures continue. One ends.

The taxonomy

Three families of risk.
Almost all of it, us.

Ord groups existential risks into three categories. The natural ones — the ones humanity didn't invent — are vanishingly small. The risks that dominate this century come from our own technology, especially the ones we haven't fully built yet.

01
Natural
Risks we didn't make
Total: ~1 in 10,000

Risks the universe has been running for billions of years. The fact that we're still here is itself strong evidence they're rare on this timescale.

Asteroid or comet impact
~1 in 1,000,000
Supervolcanic eruption
~1 in 10,000
Stellar explosion (nearby)
~1 in 1,000,000,000
02
Anthropogenic
Risks we've already built
Total: ~1 in 30

The risks the 20th century introduced and the 21st has not yet defused. Real, present, and almost entirely a function of human choice.

Nuclear war
~1 in 1,000
Climate change
~1 in 1,000
Other environmental damage
~1 in 1,000
03
Future / unaligned
Risks we're about to build
Total: ~1 in 6

The technologies of the 21st century. The bulk of Ord's estimated risk lives here — and the single largest line is artificial intelligence we cannot align with human values.

Unaligned artificial intelligence
~1 in 10
Engineered pandemic
~1 in 30
Naturally occurring pandemic
~1 in 10,000
Other unforeseen / "unknown unknown"
~1 in 30
Other anthropogenic (nanotech, etc.)
~1 in 50

All probabilities reflect Ord's estimated chance of each risk causing an existential catastrophe within the next 100 years. They are uncertain by their nature; he treats them as best-available order-of-magnitude judgments. Bars are scaled relative to the largest line in each family for legibility, not absolute probability.


Why now

For 200,000 years,
the risk was background.
For about 75, it hasn't been.

Ord's argument: humanity passed an inflection in 1945. With the bomb, for the first time, our own technology could plausibly end us. The Precipice begins there.

Every decade since has compounded the bet. Nuclear arsenals, then biotechnology, then advanced AI — each capable of wiping the slate, each arriving faster than our institutions know how to govern it.

The Precipice isn't a forecast of doom. It's a description of a narrow stretch of history we have to walk through carefully. If we make it across, the upside is staggering: a future of millions of years and possibly trillions of lives. If we don't, there is no across.


What to do

The work of the
Long Reflection.

01
Reach safety
Get the existential-risk rate down. Treat AI alignment, biosecurity, and nuclear de-escalation as civilizational priorities, not academic specialties.
02
The Long Reflection
Once safe, take the time we need to figure out what a flourishing long-term human future actually looks like — before locking in any irreversible direction.
03
Live into it
Become the kind of people, and build the kind of culture, that could plausibly steward such a future. That's where the Era of the Soul work meets the Precipice.

Walk the
Precipice awake.

The risks are real. The window is also real. The work is to get to the other side — and to be worth what we find there.