5.1 This can't go on

The world economy has grown by roughly 2% a year for about two centuries. To anyone alive today, that feels like the normal state of things: each generation a bit richer than the last.

Holden Karnofsky argues that it is anything but normal. For most of human history, growth was close to zero. And it is easy to show that 2% a year cannot go on for very long, in historical terms.

8,200 years sounds like a long time, but it is shorter than the time since humans began farming. The point isn't that growth will stop in 8,200 years. It is that the past two centuries are an anomaly, and anomalies end. Karnofsky sees three broad possibilities:

• Stagnation. Growth slows to a crawl and stays there.
• Explosion. Growth speeds up dramatically before hitting physical limits.
• Collapse. War, pandemics or other catastrophes knock civilisation backwards.

Whichever happens, we are probably living through an unusual and unstable period, not a calm stretch of business as usual.

Karnofsky's wider argument is that advanced AI is the most likely route to the "explosion" scenario. Systems that could do the work of scientists and engineers could speed up research, including research on AI itself. This is why the rest of this unit is about AI.

A caution. Economists point out that growth is measured in value, not atoms, so the comparison can be challenged. Growth could come from better ideas rather than more stuff. The argument's narrower claim is harder to dismiss: today's growth rates are historically recent, and assuming they will simply continue is itself a strong bet.