In the 19th century, economist William Jevons noticed something strange: when steam engines became more efficient and started using less coal, total coal consumption didn't fall — it skyrocketed. Because cheap energy became profitable to use absolutely everywhere.
The exact same thing is happening now with LLMs and AI agents, only instead of coal it's our brains and time. Coding, basic analytics, research — all of it became many times faster and cheaper. You'd think: finished a week's work in an hour — close the laptop and go for a walk.
But that's not how it works.
Because cognitive labor got radically cheaper, demand for it became practically infinite. If a team used to ship one feature per month, now they're expected to deliver ten tested hypotheses per week. We don't necessarily sit at computers 16 hours a day, but the density of our work grows exponentially. We generate more output, manage a bunch of agents, and launch production cycles that used to cost millions of dollars.
This is exactly the Jevons Paradox flywheel that fuels the subjective feeling of approaching singularity. When the cost of a single intellectual iteration approaches zero, the number of those iterations around us approaches infinity. So there won't be less work — it will just be different, and everything will happen frighteningly fast.