Rise of the machines: why artificial intelligence will power Google’s ‘core’

Given a large enough set of data to work with, computers can develop their own strategies for noticing patterns and making judgements. That’s the main idea behind machine learning, a subset of artificial intelligence, which is already helping Google, Microsoft, Facebook, and other companies to translate languages, filter e-mail, and find information a user might want.

But at a quarterly earnings call on Thursday for Alphabet, Google’s parent company, Google CEO Sundar Pichai reported that machine learning is about to change everything about the way tech companies tackle big problems. 

“Machine learning is a core, transformative way by which we’re rethinking everything we’re doing,” Mr. Pichai said during the call. “We’re thoughtfully applying it across all our products, be it search, ads, YouTube, or Play ... We’re in the early days, but you’ll see us in a systematic way think about how we can apply machine learning to all these areas.”

Google isn’t the only company that’s focusing in on machine learning. Machine learning allows Microsoft-owned Skype Translator to translate between spoken languages in near real time, since the system learns context and nuance through hearing words spoken.

Earlier this year, Facebook’s AI Research division open-sourced machine learning code that could lead to better image and video recognition as well as intelligent ad placement.

The push toward machine learning is being driven by two factors. First, new data is being produced at a staggering rate: in 2012, humans created 2.8 trillion gigabytes of new information, and research firm IDC estimates that the total amount of information in the world now doubles about every 18 months. That’s far too much data for any team of experts to make use of, but it’s perfect for machines: the more information computers have to crunch through, the better their algorithms become at understanding relationships between different bits of information.

Second, teams at universities and companies have made some pretty big breakthroughs in artificial intelligence in the past few years. In 2009, a group led by AI pioneer Geoff Hinton cut the number of errors in automatic speech recognition engines by 25 percent, a feat Alphabet chairman Eric Schmidt referred to as “the equivalent of about ten years of research all at once.” And earlier this year, Dr. Hinton (who was hired by Google in 2013) said that the company had developed “thought vectors” that could allow computers to process language in much the same way humans do.

Pichai’s announcement on Thursday indicates that Google is expecting machine learning to become the core of all sorts of areas, including search (predicting what information people will want to find, even before they’ve asked for it) and advertising (judging what products and services people will find appealing based on more than just what websites they’re visiting).

Artificial intelligence is no longer a far-future sci-fi concept, nor an afterthought bolted on to a few digital products – instead, guided by human experts, machines themselves are beginning to develop strategies to tackle big challenges in technology.