Google uses DeepMind-Powered AI to manage power usage
Internet giant Google started using Artificial Intelligence technology of its subsidiary DeepMind to reduce the energy consumption of its data centers. DeepMind's Co-Founder, Demis Hassabis, stated Alphabet Inc. put the DeepMind AI system in charge of its data centers to manipulate servers, cooling systems, and other equipments to reduce power consumption. High energy consumption and cooling systems have been a headache for technology companies.
DeepMind is a UK-based artificial intelligence startup that specializes in machine learning, systems neuroscience, and advanced algorithms. Demis Hassabis, a neuroscientist; Mustafa Suleyman, and Shane Legg founded DeepMind in 2012. DeepMind aimed to design computers that can think like humans; it develops gaming and e-commerce technologies. Google acquired the two-year-old AI startup for $650 million in one of the biggest European acquisitions in 2014.
Google says, "Specifically, our data centers use only 50% of the energy of most other data centers. In addition to reducing our impact on the environment, our efficient data center designs have saved us over a billion dollars to date."
The result of Google's AI experiment was a 40% cut in the power requirement to cool down servers. Google described the result as a "phenomenal step forward"; the 40% reduction resulted in a 15% cut in overall energy consumption. According to Demis Hassabis, the AI technology used in data centers is similar to DeepMind's software, which taught itself to play Atari games.
In 2014, Google used 4,402,836 MWh of power; which is equal to the power consumed by at least 366,903 households in the US. Google said the 15% reduction in overall energy consumption would save "hundreds of millions of dollars" in the future.
Hassabis stated the exclusively made neural networks control nearly 120 variables, including fans, cooling systems, windows, etc. in the data centers. The AI technology cools systems by analyzing sensors' data, including pump speeds and temperatures in the server racks. DeepMind's next step would be identifying those places that require new data to calculate efficiencies and install sensors there.
DeepMind stated it wouldn't stop with just Google's data centers. The company added the "algorithm is a general-purpose framework" that helps to understand complex dynamics. They intend to apply it to several other challenges beyond data centre environments in the future. It said by using Artificial Intelligence to reduce energy consumption; one of the biggest challenges - climate change - can be addressed.
DeepMind wrote in its blogpost: "From smartphone assistants to image recognition and translation, machine learning already helps us in our everyday lives. But it can also help us to tackle some of the world's most challenging physical problems -- such as energy consumption."
