From the first stone tool to the transistor took three million years. From the transistor to machines that settle open problems in mathematics took eighty.
The gaps between inventions keep shrinking. A sequence whose gaps shrink by a steady ratio does not run on for ever. It closes on a line. No one knows whether history will hold that ratio. We do know what stands on the line.
The last invention
A machine that can discover. In 1965 I. J. Good called it “the last invention that man need ever make”.1 After it, every other invention can come from the machine.
What we are building
That machine. A system that finds what no one has written down, and proves it. The proof matters as much as the finding: it is what lets a result be trusted, and the next one be built on it.
Discovery a hundred times faster. A century of science, every year.
What it changes
It reaches wherever an answer can be checked: mathematics first, then far more of physics, chemistry, biology and engineering than it first appears. Questions that would have outlived the people who asked them get answered in time to matter.
Where we are
We began in mathematics, where a result is right or wrong and a computer can tell which. The results so far are stepping stones. Superintelligence will be recognised the same way, by its record: discoveries no person made, each one verified, each making the next cheaper.
- 1. I. J. Good, “Speculations Concerning the First Ultraintelligent Machine”, Advances in Computers 6, 1965.