In the raging discussion on the potential for computers to pick up our jobs we see day after day new examples of human capabilities that are acquired by computers. Just few days ago I reported on a computer that can perform autonomous surgery.
Today is a news about a computer that can perform chemical experiment to find new alloys.
A joint team at Los Alamos and State Key Labs in China have found a way to combine machine learning, massive supercomputer processing power to perform experiment and speed up the discovery of new materials with sought after properties.
Clearly one may claim that here we are looking a brute force, not at intelligence but the claim could be refuted by saying that we, human beings, are learning from experience. And intelligence is about to be able to learn, and possibly learn fast.
This is actually what is described in the Nature paper. The researchers have created a software that allows a (super)computer to learn from experiment, so that it can restrict the slate of options and converge on what is most promising.
Alloys are becoming more and more complex, along with requests of specific characteristics, like memory shape alloys that have the capability to regain their shape after an induced deformation. Researchers are reporting that a computer with learning capability can cut in half the research time for new alloys, shortening the time to market.
This is very interesting as we are moving towards smart materials and nano-tech fabrication capability. They will play a significant role in next decade fabrication, becoming an important component of Industry 4.0.
As I finished writing this post I saw another news: a law company has hired Al Ross to manage their bankruptcy practice. Nothing strange, were it not that Al is a software attorney, a service provided by IBM through Watson. The number of jobs that can be executed by smart systems keeps growing... at an alarming pace.