Google AutoML Writes Code Faster Than Humans – AI Beats Programmers

Google's AutoML system recently produced a series of machine‑learning code whose efficiency surpasses that of the researchers who created it, striking another blow to the notion of human superiority as robot "students" become masters of self‑replication.

The team introduced a machine‑learning software capable of creating self‑learning code; it runs thousands of simulations to identify which parts of the code can be improved, iterating the process until the target performance is reached.

This serves as a vivid illustration of the "infinite monkey theorem"—instead of a monkey typing Shakespeare, Google built a machine that can self‑replicate programming, achieving in hours what human programmers would need weeks or months to accomplish.

In programming machine‑learning systems, AutoML far outperforms its creators. On an image‑recognition task it achieved a record 82% accuracy.

Even in complex AI tasks, its self‑generated code is superior: it can label multiple points in images with 42% accuracy, compared to 39% for software written by humans.

Google announced AutoML only five months ago; given its ability to produce a better AI system so quickly, the results expected in the coming year are highly anticipated.