Using technology developed by one of their recent acquisitions, Google claim to have cut down energy usage at their large data centres.
In 2014, Google acquired DeepMind, a then-small company focused on artificial intelligence and machine learning, for £400 million. Now, they say, by employing artificial intelligence to manage their data centres far more efficiently than humans could, they have managed to reduce energy usage at the sites by as much as 15%.
Google have been working on machine learning technology for some time now, employing it to varying degrees in various aspects of its work from photo recognition to search results.
Now, in what will certainly be hailed as a major step for the field as a whole, it has been employed in a very directly tangible way, making use of the technology’s ability to perform vast and complex calculations much better and more efficiently than humans could.
DeepMind’s co-founder, British entrepreneur Mustafa Suleyman, explained: “It’s one of those perfect examples of a setting where humans have a really good intuition they’ve developed over time but the machine learning algorithm has so much more data that describes real-world conditions.
“It’s much more than any human has ever been able to experience, and it’s able to learn from all sorts of niche little edge cases seen in the data that a human wouldn’t be able to identify. So it’s able to tune the settings much more subtly and much more accurately.”
Suleyman described the DeepMind’s major advantage as it being able to “tweak all of the knobs simultaneously”, being able to accurately predict potential usage, and to tweak the data centre’s power consumption accordingly. The majority of the reduction in energy use came from intelligent adjustment of cooling technology, which was shown to use 40% less power with the help of DeepMind.
Reducing energy use in Google’s data centres, and, in the future across more of its faculties, is important due to the growing focus placed on energy consumption and carbon emission from the digital world generally.
Data centres across the world are, according to one recent study, responsible for around 2% of the world’s carbon emissions and Google, by their own admission, is responsible for 0.01% of all of the electricity used in the world.
“Putting a dent in that,” Suleyman said, “benefits the world in general.”
In 2014, Google said, they “reported a carbon footprint of 2.49 million metric tons of carbon dioxide equivalent (CO2e)”. This is set to improve, they say, given that in the same year, 37% of the electricity they used came from renewable sources – a 2% increase from the year before.
Salesman spoke positively about the future of this use of DeepMind’s technology, saying: “I really think this is just the beginning. There are lots more opportunities to find efficiencies in data centre infrastructure.”
He spoke of his desire to ultimately roll out the technology to other, non-Google partnered companies, explaining that “one of the most exciting things is that the kind of algorithms we develop are inherently general”, allowing “the same machine learning system [to] perform well in a wide variety of environments.”
