Korean grandmaster Lee Sedo suffered defeat in Go – a 2,500 year old game exponentially more complicated than chess – at the hands of Google’s Artificial Intelligence program in March.
That same technology is now conserving power at a handful of Google data centers, slashing the energy bills for cooling by 40 percent this spring.
The achievement opens the horizon for energy efficiency gains in manufacturing plants, power plants, and energy grids, all thanks to an artificially intelligent system that can master any pattern, whether on a game board or in a heating and cooling system.
DeepMind uses "deep learning," recognizing patterns in large data sets, to analyze information from thousands of sensors measuring temperature and power usage in each data center. Google’s data centers support Google Search, Gmail, and Youtube, generating high levels of heat that must be removed in order to keep the servers running.
The 40 percent drop in cooling costs corresponds to a 15 percent improvement in power usage efficiency, Google said in a statement.
"We're really thrilled about the environmental impact," Mustafa Suleyman, who leads applied AI at Google DeepMind, told CNN Money.
The AI has also spotted gaps in its coverage and will instruct Google on where to install more sensors to further increase its effectiveness.
All of Google’s data centers will employ the system by the end of the year. The energy savings potential in data centers alone is huge, since according to some estimates, they account for 2 percent of global greenhouse gas emissions.
And artificial intelligence can bring about even more energy-saving in other areas as well.
DeepMind will release the details of the system in a white paper within the next six weeks, so that manufacturing facilities and even national energy grids can also take advantage of the energy-saving system, the BBC reported.