How Google is getting smarter with artificial intelligence
Google (GOOGL) kicked off its developer conference Wednesday by highlighting its progress with artificial intelligence.
CEO Sundar Pichai opened and closed the closely watched event at its Mountain View, Calif., headquarters by outlining the Internet giant's research in AI and machine learning. Among the applications Google sees for this technology is teaching computers how to "see" what's in a person's photo, as well as understanding what users are asking a software program to do, rather than simply deciphering spoken language.
All of that baseline research, which companies such as Facebook (FB), IBM (IBM), Microsoft (MSFT) and China's Baidu (BIDU) are also pursuing, today is already being used in Google products in some way, he said. Notably, Pichai explained, Google's focus on AI shows its efforts to move away from its search engine's reliance on links and instead delivering information based on where users and what they have asked for in the past.
Given the proliferation of mobile devices with tiny screens and keyboards, as well as a realization that connected homes, cars and wearables mean that screens become less of a focal point for interacting with computers, Google is hopping on an established trend. Still, the company has one of the strongest machine learning and AI research groups in the world. It also has vast amounts of user data that it can tap to strengthen its AI technology. That's a potent combination that could move the tech forward.
For example, Pichai showed off a search interface that lets people ask Google, "What's playing tonight?" and returned cards that listed three movies local theaters. In this case, the movies were all R-rated, so Pichai continued the conversation, telling Google he wanted a film to which he could bring the kids. That elicited listings for three family-friendly movies.
The executive completed the exercise by using Google Assistant on his phone to buy tickets to a showing of "The Jungle Book."
Such applications resemble Facebook's M efforts, only in this case there are no people on the back-end making sense of queries Facebook's AI can't answer. This is a natural evolution of Google's focus on deriving context about you based on your history, emails and location.
Google's conversational interface, which like Apple's Siri or Microsoft's Cortana personal digital assistant, has a female voice, is used in a "smart" speaker called Google Home that could be rolled out later this year that lets users control connected home products, answer questions and play music.
Unlike Amazon's Echo smart speaker system, the Home lets a user play music across multiple rooms and answers more complex search queries using Google's combination of machine learning. It also appears to be able to offer notifications through the speaker, alerting a user to flight delay or an upcoming appointment, among other things.
Google Home doesn't have a launch date yet, but as a physical embodiment of a personal assistant it could become a compelling addition to the home. It's also a slap in the face to Nest, the company that Google purchased for $3.2 billion in 2014 and that was supposed to be the core of its smart-home platform.
The other big consumer news emerging from the event was the creation of Daydream, a virtual reality platform that includes a design for a headset and controller, along with an app store, that lets users to buy VR games.
Until now, Google's dedicated VR efforts have focused on Samsung VR hardware at the high end and a cardboard-sleeve that people slid their phone into to watch VR content on the low end. With Daydream, Google is creating a full VR ecosystem that eventually could compete with Facebook's Oculus, Samsung Gear and HTC's Hive.
Daydream has the potential to usher in cheaper virtual reality because it will work on a range of Android smartphones. These phones will need a specific set of sensors and screens to run Daydream-compatible games and content, but Samsung, Asus, LG and other Android phone companies have already signed up. This could bring VR to the masses, although questions remain about quality and battery life.
Google also shared details on tech that could presage some great user experiences on the way for Android users.
For example, Instant Apps for Android takes the concept of installing an app and tosses it out the window. Now when you get a link to content that requires you to open an app (LinkedIn comes to mind), Android will actually grab the content from that part of an app so you can view it without installing the app. This is great for those times when you just want to see content or do something like pay for parking in a strange city, without installing another app.