Whether it's the launch of Google's new photo service, Google Now or the new Android, machine learning appears to be an emerging common thread.
Google rolled out Google Photos and the differentiator is going to be search and machine learning to categorize them.
The pitch with Google Photo is that the search giant will create a home for photos and videos and use machine learning to "make memories not manage them," said Anil Sabharwal, director of Google Photos.
All of this machine learning computing power will rely on Google's computing and cloud infrastructure. Google isn't the only one in the machine learning game---IBM, Microsoft and a bevy of others play in the space---but the search giant has the reach to make the technology mainstream even if it resides in the background.
The other machine learning piece is what Google is doing with Google Now, which is gunning to be more current. Google is hoping to engage customers with a voice request for a name.
Google Now is also adding an On Tap function that serves as a help me button. The natural language processing is aimed at giving users context and answers quickly.
You can see where Google is heading. "We have built up a natural language processing engine, but we have also built up this powerful context engine, and we understand more than 100 million places," said Aparna Chennapragada, director of Google Now.
All of this is possible because Google has improved its speech recognition accuracy from a 23 percent word error in 2013 rate to 8 percent today.
"We have the best investment in machine learning over the past many years," said Sundar Pinchai, senior vice president at Google. Machine learning and deep neural nets aren't exactly new, but Pinchai made the case that Google has spent its time training.
"What looked like a simple query, we understood voice, we're doing natural language processing," he said. "The reason we are able to do all of this is because of the investments we've made in machine learning."