Bloomberg reported that the technology giant was able to use the artificial intelligence to reduce power consumption in its data centres. According to a blog post, DeepMind was able to help reduce the amount of energy used for cooling by up to 40 per cent, which equates to around an overall 15 per cent reduction.
This is impressive considering Google said it used around 4,402,836 MWh (megawatt-hours) of electricity in 2014. For perspective, that’s around what 366,903 US homes use yearly.
DeepMind is a British firm that was acquired by Google in 2014. The artificial intelligence company builds neural networks that run off of algorithms, basically creating machines that can adapt in a way similar to how humans learn.
Google has been using machine learning for a couple of years in order to run the data centres more efficiently, but implemented DeepMind a few months ago. By utilising the data that had been accrued over those couple of years, engineers were able to train the neural networks to recognise the average Power Usage Effectiveness (PUE), or the ration between total building energy use and IT energy use. By using this, and predicting what future numbers would look like, the DeepMind was able to calculate the proper actions.
“It controls about 120 variables in the data centres. The fans and the cooling systems and so on, and windows and other things,” DeepMind Co-Founder Demis Hassabis said.
This makes sense if you look at other DeepMind applications. It’s been programmed to get the high score in games, so why not get the “high score” in energy usage? The algorithms that it utilises can also be used for even more applications in the future. Even Google said it’s going to use the neural network more “broadly” throughout the company.