Alpha Dog Technology New Uses: Help Google Reduce Electricity

Google's deepMind co-founder Demis Hassabis (Demis Hassabis) recently in New York's artificial intelligence conference, said Google is using DeepMind technology from the data center to optimize the energy consumption, thus saving a lot of energy costs.

In recent months, Google is using DeepMind's artificial intelligence system to control some parts of the data center to optimize the energy consumption of servers and refrigeration equipment. Previously, DeepMind's technology was used to teach computers to play Atari games.

Hassabis said the technology has reduced the energy consumption of Google's data centers by a few percentage points. "From the cost point of view, this brings great savings, but also conducive to environmental protection."

Google said that this optimization brought about 15% of the energy efficiency improvement. Energy efficiency indicators will be computer power consumption and refrigeration systems and other infrastructure power consumption compared.

Google said that in 2014 the electricity consumption of 4402836 MWh, equivalent to 366,903 US households in the average annual power consumption. A large part of these power consumption comes from the data center, and the data center is the basis of Google web services and mobile applications.

Reducing power consumption by a few percent will be a huge financial help for Google. According to the US Energy Information Administration, in the United States, electricity prices are usually 25 to 40 dollars per megawatt hour. Therefore, if the data center energy consumption by 10%, then in a few years time Google can save hundreds of millions of dollars. Google in April to 400 million pounds of the price of the acquisition of DeepMind, when this is equivalent to more than 600 million US dollars.

Google has also used machine learning technology in the data center. In 2014, Google said it was using a neural network system to predict energy changes over time, making it more efficient to deploy equipment.

DeepMind's work on this basis goes further. The company's software can adjust the operation of the data center equipment, improve energy efficiency. "This controls about 120 variables in the data center, including fans, cooling systems, and windows and other devices," Hassabis said.

He also said that this is only the beginning of the project. At present, DeepMind knows that this method is effective, but also began to understand what information it lacks of artificial intelligence system. As a result, DeepMind may ask Google to deploy new sensors in the data center to help the software get more information.