Google recently leveraged its DeepMind artificial intelligence (AI) system to save some money on its energy bills. According to DeepMind co-founder Demis Hassabis, DeepMind has been put to work in Google’s data centers and made them more efficient.
Originally reported by Bloomberg, DeepMind helped lower power consumption by “manipulating computer servers and related equipment like cooling systems.” Hassabis said that it accomplished this by using some of the same techniques it used when DeepMind learned to play Atari games back in February 2015, by digesting data and finding patterns that it can utilize.
When all was said and done, the system’s power usage efficiency (PUE) improved by 15%. It’s important to note that PUE measures the power consumption of computing equipment, independent of the other systems involved with a data center. It currently controls 120 variables in the data centers, Hassabis said.
According to the 2014 Google Green Report, Google’s data centers already use 50% of of the energy of typical data centers. However, as Bloomberg noted, Google still used 4,402,836 MWh of electricity in 2014, which is about the same amount used by 370,000 homes in the US in a year.
SEE: How Google’s DeepMind beat the game of Go, which is even more complex than chess (TechRepublic)
The effort is merely the latest in a series of steps the company has taken toward sustainability in its data centers. In March 2016, at the GCP Next conference in San Francisco, Google’s Joe Kava explained some of the sustainability efforts Google was taking and expressed a desire for research firm Gartner to include sustainability as a measurement in its Magic Quadrant report.
At the time, Kava said the key tenets of Google’s culture were “ownership, sustainability, and innovation,” also noting that the company had invested $2 billion in renewable energy.
The 15% figure may not seem like much, but considering just how much energy the search giant uses per year, even a 15% improvement could mean savings of millions of dollars. This, essentially, could end up paying for Google’s acquisition of DeepMind, which is made back in 2014 for more than $500 million.
DeepMind was founded by Demis Hassabis, Shane Legg, and Mustafa Suleyman. The company’s machine learning system is most known for beating a champion player of the board game Go, and even tried creating cards for Magic the Gathering as well. However, this latest application could be evidence of the daily usefulness we will see from machine learning in the future.
The 3 big takeaways for TechRepublic readers
- Google’s DeepMind AI controls over 120 aspects of its data centers, and made the power consumption of the computer equipment 15% more efficient.
- With the massive amount of power Google consumes, the additional efficiency could save the company massive amounts of cash in the future.
- The application of DeepMind to energy use could prove how machine learning and AI will continue to impact our daily lives in the future.