Google’s AutoLM Program Has Created its Own Artificial Intelligence
In recent years, Google has gotten heavily involved in the world of artificial intelligence. The company made headlines when it programmed an AI that defeated the Go world champion in a best of five rounds tournament. It was an impressive demonstration of Google’s machine learning programs. However, the company is taking things a step farther by working on an AI that programs other AIs.
At a recent event in Tokyo, Google Brain Senior Fellow, Jeff Dean, spoke about the future of the company’s AI program and one of the areas he touched upon was AutoML.
“What if we could automate the process of machine learning,” Dean asked?
Most AIs are created using neural networks designed to simulate the way the human brain works. This process, called machine learning, can be used to train computers to recognize patterns in various forms of information. It’s proven effective, but it requires large amounts of data and human input to ensure it is on the right track.
However, Dean has said that Google’s AutoML program has created a new neural network without any human input. Beyond that, the AI-created neural network is apparently much more powerful than any of the human-created ones.
As proof of this, Dean shared a slide showing the aforementioned AI being used to recognize images. This is a fairly common use for an AI and the one created by the AutoML program managed to score an accuracy rating of 82 percent which is higher than that achieved by human-created AI.
This feat represents a major advancement in the field of artificial intelligence, but we can’t hope, but wonder about the long-term consequences of this sort of technology. It’s a bit too early to start worrying about Skynet, but the implications of this technology are both fascinating and more than a little unnerving.
Eric is an avid tech junkie, gamer, and comic fan. When he's not working on his PC, you'll find him at your local comic book shop.