There weren’t many shocks at Google’s annual developer conference, but the company did remind us of one area where it’s blowing the competition away.
Teaching machines how to think.
Arguably the most exciting announcement at Google I/O was the introduction of Google on Tap: a more advanced version of Google Now that can actually anticipate what you need before you ask.
It does this by assessing the information on your phone to pull up relevant bits of data, such as locations, definitions, reviews, and more. And it accomplishes this through machine learning.
Machine learning refers to computer algorithms for learning to do things. A lecture from Stanford University’s Rob Schapire explains it quite well: “In other words, the goal is to devise learning algorithms that do the learning automatically without human intervention or assistance.”
Based on early impressions from the media, it sounds like even the unfinished version of Google on Tap is pretty mind-blowing. Google claims you can speak to it just like you would a friend — the natural language processing is supposedly that advanced. All signs are pointing to a super powerful virtual assistant coming to Android, and a big part of that is Google’s machine learning and artificial intelligence division.
Microsoft has been a pioneer in this area as of late too — its Skype voice translator, which can translate speech in real time — showcased how impressive the company’s machine learning technology has become. Microsoft’s virtual assistant, Cortana, also learns more about you and your habits the more you use it.
Apple’s Siri has gotten smarter and faster over time , but it’s not contextual like Google on Tap, or even the current version of Google Now and Cortana. It’s quick, and sufficiently accurate most of the time, but it can’t anticipate my needs or learn who I am.
“I think Google is way out in front of everybody, including Microsoft, but probably furthest out in front of Apple,” Van Baker, a Gartner research analyst, said to Business Insider when asked about Google’s accomplishments in machine learning. “Microsoft is going in that direction, and I think calling [Apple] behind is very fair.”
A big part of that is Google’s powerful search engine — machine learning and understanding patterns are a big part of why Google’s search results are so accurate and timely. So, it makes sense that the biggest search company in the world would be miles ahead of the competition in terms of machine learning.
Apple isn’t in the search business — and it doesn’t need to be — so machine learning isn’t as big of a priority.
Machine learning is a part of nearly all of Google’s core products: Google Maps, self-driving cars, and even ads. Matthew Zeiler, the CEO of Clarifai that previously interned at Google, even told Wired last year that Google isn’t really a search company, but a machine learning company.
“Everything in the company is really driven by machine learning,” he told Wired.
Google is applying that concept to it’s new Photos app, too, which will use machine learning to make your image albums just as searchable as web links.
The reason we are able to do all of this is because of the investments we have made in machine learning,” Google’s senior vice president of product, Sundar Pichai, said on stage at the event. “Machine learning is what helps us answer the question ‘What does a tree frog look like?’ from millions of images around the world.”