Making sense of Google CEO Sundar Pichai's plan to move every direction at once

Google CEO Sundar Pichai has laid out a plan that could, if successful, help parent company Alphabet maintain double-digit growth in sales and profit: Use Google's massive scope to build the best machine learning systems in the world. Then use that expertise to build apps and services that consumers can't live without.

In the process, Google could once again reshape how consumers use the Internet.

Pichai kicked off a sprawling two-hour keynote presentation on Wednesday at Google I/O 2017, the company's annual conference for programmers, by throwing out some impressive stats. The crowd was particularly impressed when Pichai said 2 billion devices were now running Android. The company also announced that Google Drive, the company's online storage service, now tops 800 million users a month. Google Photos, a way to organize and share photos online, is over 500 million.

But these billions of users are just a prelude. What Pichai is really excited about is machine learning -- software that gets smarter over time by munging reams and reams of data.

"Thanks to advances in deep learning, we're able to make images, photos and videos useful" for finding things, he said. "Speech and vision are becoming as important to computing as the keyboard."

Google's ability "to understand images and video has profound implications" for all of what it does, he said.

To that end, he introduced Google Lens, a product for recognizing and locating images within videos and photos, and noted that Google Home -- the company's competitor to Amazon's popular Echo -- can distinguish between different voices. A slide behind him also noted that Google's error rate in voice recognition is now just under 5 percent, down from 8.5 percent in July, 2016, according to one slide behind Pichai.

But the plan is bigger than the company's own services. Pichai also wants to make it easier for developers to use these artificial intelligence advances in their own apps -- and this was a conference for developers, after all.

"We want it to be possible for hundreds of thousands of developers to use machine learning," Pichai said.

Google will make this intelligence available through its cloud computing service, made more powerful by an update to a new kind of chip called the Tensor Processing Unit, or TPU. Google has been working on this project for several years, and while it won't sell the chips, it will allow any developer to access the power they provide through Google's cloud services.

If his plan succeeds, Google technology will be able to provide answers to a broad swath of consumer questions and solve a growing number of their problems.

The most valuable tech firms — powerful rivals all — are attacking similar problems.

Amazon has built a personal assistant that beat Google to market with hands-free calling and product-image recognition, and makes its own AI services available through its market-leading cloud service, AWS. Microsoft is also in the game with its personal assistant, Cortana, and machine learning services available through its Azure cloud.

Apple has Siri and is surely working on bigger things in the background. Uber is pursuing self-driving cars and doing the the AI work necessary for that initiative. Facebook, IBM, Salesforce, even Snap -- all of them are in the game. In fact, you can't listen to a start-up pitch in Silicon Valley today without hearing the words "machine learning."

But Google has a much broader scope than any of them. No other company has a market-leading mobile platform, online video service, web browser, and multiple online services with hundreds of millions of users.

In the machine learning game, the more data you have, the faster your programs learn. That's what Pichai and Google were really showing off on Wednesday. It may have seemed like a disorganized sprawl, but that diversity is Google's biggest strength, and could give it the edge it needs to win the next generation of computing.