With the ability to help robots learn on their own, Google plans to teach robots how to “teach” each other, thereby creating higher-level artificial intelligences (AIs).
Tech giant Google recently announced a huge decision related to AI (artificial intelligence). Accordingly, they are aggressively promoting self-learning AI, then making them build a more optimal AI system on their own. Simply put, humans are teaching robots how to “teach” each other.
In the near future, robots will produce the next generation robot by themselves.
Specifically, Google said that it is designing an artificial nervous system that simulates the activity of the brain. And their new technology – AutoML, will make this system work more efficiently and intensely.
” The mechanism here is that we create a group of neural networks, and iterate over these same networks until we find the best one .” – Sundar Pichai, CEO of Google said.
This process is called “reinforcement learning,” where computers associate trial and error with each other, in the same way that humans train dogs. But the difference is that, AutoML will recognize which system is the most optimal and intelligent.
However, doing them takes a lot of energy, and Google is having to solve that with AutoML.
“We hope AutoML can replace a small fraction of today’s experts, and in the next few years hundreds of thousands of people, to create a separate network just for them without going through a human.” – Pichai shared.
AutoML is in its infancy right now, but Google says AI platforms will quickly become smarter thanks to it. In the near future, we’ll see their effects on Google’s smartphone apps.