At last year's Google I/O developer conference, CEO Sundar Pichai described how the world was shifting from mobile-first to one led by artificial intelligence (AI). This year, the company is focusing on "making AI work for everyone," he said.

Now in its final day, this year's event has highlighted Google's increasing use of AI and machine learning for everything from email to job-hunting to healthcare. The company's latest initiatives even include one that uses machine learning to design better machine learning.

During the event, Google also unveiled the second generation of its tensor processing units (TPUs), chips custom-designed for machine learning. The new cloud TPUs will be made available to developers and researchers via Google's Compute Engine, with an alpha program for users set to kick off sometime in the future.

Immersive Computing for Everyone

Google also continues to work on advancing its capabilities in augmented reality (AR) and virtual reality (VR), and investing in platforms that will make them more accessible to a wider audience, according to VR/AR vice president Clay Bavor.

"In order for immersive computing to be 'for everyone,' developers need to build great apps and experiences," Bavor wrote yesterday in a blog post. "We're working on tools and tech to help them do that."

A new tool called Instant Preview, for example, is designed to let developers using VR headsets see the results of programming changes they make "in seconds, not minutes," Bavor said.

Another tool coming soon, Seurat, is aimed at helping mobile developers achieve high-fidelity, real-time scenes and graphics that are equal to or better than those on desktops, he said. And this summer, Google will launch Chrome VR, which will enable users of the mobile VR platform Daydream to browse the Web in virtual reality. The tech giant also released an experimental build of its open-source Chromium browser with an API for augmented reality.

Smarter, Faster Gmail Replies

More commonly used Google apps are also becoming increasingly smart. For instance, the company announced Wednesday that its Smart Reply feature for Gmail will begin rolling out over the next two weeks for Android and iOS mobile devices. It will be available first in English, with Spanish rolling out in the coming weeks and more languages to follow.

Smart Reply offers three response suggestions based on the content of an individual email message. A user can either tap one of the options to send it immediately, or select a response to edit it before sending.

"Smart Reply utilizes machine learning to give you better responses the more you use it," Gmail software engineer Greg Bullock wrote on the G Suite blog. "So if you're more of a 'thanks!' than a 'thanks.' person, we'll suggest the response that's, well, more you!"

Over the coming weeks, Google also plans to launch a new Google for Jobs feature that will incorporate AI into its Search function to make it easier for employers and job seekers to make better-focused matches.

During today's final day of Google I/O, the schedule includes sessions on how to push the boundaries of machine learning, how to apply "built-in hacks of conversation" to voice-based user interfaces, and how to use behavioral insights to boost user retention.

The 2017 Google I/O developer conference -- or festival, as the company likes to call it -- has been running May 17 - 19, at the Shoreline Amphitheatre in Mountain View, California.