Google introduced its AutoML project back in May. As an AI capable in creating more AIs, the system has been able to produce a series of machine-learning codes with higher rates of efficiency than those made by the researchers themselves.
While AI is still young and developing, this advancement has concluded that AI can indeed self-replicate, or 'reproduce'.
AutoML was initially developed as a solution to the lack of talents in AI programming, even at Google.
When Google created AutoML, the company taught the machine learning software to create self-learning code. Then Google allowed the system to tun thousands of simulations to determine which areas of the code can be improved or changed.
When the AI has learned how to do things properly, the system can do in hours what takes the human programmers weeks or months.
As a result, AutoML became a machine-learning system capable of coding better than the researchers who made it.
In an image recognition task for example, it reached a record high of 82 percent accuracy. Even in some of the most complex AI tasks, its self-created code is also superior to humans: it has beat human-built systems integral to autonomous robots and augmented reality.
For that task, the system created by AutoML was able to mark multiple points within an image with 43 percent accuracy compared to human-made software at 39 percent.