Artificial Intelligence: Google itself draws knowledge from games. Its … – Gadzeto Mania

The artificial intelligence is much talk and write, but rarely have the opportunity to see her in action. Google has released a video showing the progress of its artificial intelligence, which was made available a simple game. The effect is astonishing and frightening at the same time – SI Google can really learn. And therefore disturbingly effective.

Skynet Google will work?

Once upon a time I wrote about the birth of Skynet, reports of artificial intelligence seemed much exaggerated . Known even from the Terminator or Matrix scenario in which the machine pay for some rational reason against a man seemed unlikely.

Warnings emerged relatively recently. First, Stephen Hawking, and later Elon Musk began to observe that people too enthusiastic approach to the development of artificial intelligence could mean the destruction of our species . And even though he does not have to be this way, you will want to take this scenario into account (more on this topic can be found in the article “Stephen Hawking warns of artificial intelligence. This is the biggest mistake of mankind?”).

These warnings seem gain on the news, when we will look at the latest developments in Google. A year ago, the company bought the Mountain View – among other things – DeepMind. The transaction, made out $ 400 million was then widely commented, among posed by various commentators questions were repeated one thing: what Google uses algorithms bought ?

The answer is now clear – buying DeepMind and others, exploring the SI, Google Places jumped to the forefront corporate holdings – because the term begins to think better reflect reality – its own artificial intelligence .

The agent learns the same

The program, called Agent is familiar with 49 titles, classic games computer at the same time not having any experience and knowledge associated with playing. I just checked a random keystrokes, following only a general guideline, the goal is to score points.

The results turned out to be amazing – after numerous attempts Agent is not just “understand” what are the some of the games, but also master the strategies developed to allow to win. An example can be shown in the film Arkanoid – let’s see, how the process of science random motions of selecting the optimal strategy , providing the best possible result.

It should be noted that such spectacular results reached so far only in simple games , which requires mastery of elementary mechanics of the game. More complex productions, requiring, for instance. Abstract thinking, are far beyond the reach of the Agent, but Google declares that soon his face artificial intelligence 3D games.

Hope is a threat to humanity?

It is probably a matter of time until it reaches the championship also results in a more complex titles. What is next? It seems that Hawking’s warning Muska and had more sense , than you might think.

Just imagine eg. The algorithm, which is connected to the Forex market with a supply of cash, allowing learning to achieve perfection . Or connect the Agent to any stock exchange, where emotionless, precise and processing huge amounts of data the would not have any chance to people , which remain perhaps the only rely on luck. No need to Skynet, to turn the world upside down.

In our eyes are made remarkable epochal breakthrough. We live in interesting times. But with all the fascination with new technologies, some of which seem a little disturbing .