Google claims to be able to design computer chips in less than 6 hours using AI, and the whole internet is truly baffled by it – because it normally takes humans months to finish this exact procedure. So, are we breaching into risky territory or should we be thrilled about this development?

According to an article published in the Nature magazine last month, Google claims to have developed deep reinforcement learning software that can generate AI chips considerably quicker than people. And by considerably, we mean record breaking.

An excerpt from the article states:

“Working out where to place the billions of components that a modern computer chip needs can take human designers months and, despite decades of research, has defied automation. This week, however, a team from Google report a new machine learning algorithm that does the job in a fraction of the time, and is already helping design their next generation of AI processors.”

In essence, Google is developing AI design chips that can create a “floorplan” for more powerful AI systems via a complex process that requires meticulously putting components such as CPUs, GPUs, and memory cores in relative places across a silicon die. Engineers generally spend months creating these floorplans, but Google’s new AI learning system, which includes a learning algorithm trained on a dataset of 10,000 different floor plans, can develop manufacturable layouts in a fraction of the time. Truly though, what on earth is happening.

Azalea Mirhoseini, one of the authors of the paper who is spearheading Google’s head of machine learning for systems, wrote:

“Our method has been used in production to design the next generation of Google TPU […] In under six hours, our method automatically generates chip floorplans that are superior or comparable to those produced by humans in all key metrics, including power consumption, performance and chip area. To achieve this, we pose chip floorplanning as a reinforcement learning problem, and develop an edge-based graph convolutional neural network architecture capable of learning rich and transferable representations of the chip.”

This new finding, according to the article, has “huge ramifications” for the chip industry’s future, allowing corporations to test possible designs more quickly and produce specialized chips for specific purposes. This might actually be helpful considering there’s a chip shortage around the world that has ramifications spreading across the telecom industry, computer industry, gaming industry, auto industry and more.

In related AI news, machine learning is now able to create music.