It's very easy to dismiss Google I/O as just an opportunity to show off new toys and to throw a big party. Given the appearance of LCD Soundsystem at this year's closing concert, the impression is understandable. But I/O 2017 was when Google's efforts in big data, search, and device ubiquity bore fruit as it transitions toward a new focus on AI and machine learning.
Google I/O 2016 is etched in my mind for two reasons. First, it was deadly hot, dampening the fun of an outdoor conference with sweat. Second, nothing announced was actually released.
In retrospect, 2016 was a building year for Google. Products announced last year, including Google Assistant and Google Home, have now been in the wild—some, like Android Wear 2, for just a few months. This year, we heard about improvements and Google opened them to developers.
Developers can now work on Android, Chrome, the Google Assistant, and Google Home through the Actions API. Instant Apps are also now available, taking away the primary friction point of downloading and using apps. The omnipresence of the Google Assistant also means Actions developed for it will filter into many new platforms, including the iPhone.
In fact, the iPhone was mentioned with such frequency, it began to feel less like a competitor and more like another platform ripe for development. The fact that the Google Assistant will be moving into Siri territory made this abundantly clear.
This is particularly notable because Android has always been the centerpiece of the Google developer experience. True, not every attendee is an app developer, but Android has always loomed large. This year, the presence of Android wasn't necessarily diminished, but it was just one of many platforms under the Google umbrella.
As a developer conference, I/O is an opportunity to not only encourage developers to adopt new tools and new platforms, but to offer some useful feedback as well. That's not unusual, but there was some tension in the discussions this year.
At the developer keynote, Android luminaries explained how new restrictions on data, processing power, and battery power for apps are not meant to be punitive but to make Android better for everyone, including users. That's a sensible viewpoint — it might even be correct — but it felt decidedly more prescriptive than past Google sessions.
That's just a few of the new ways Google is reclaiming Android. Project Treble, described in the Android security panel, divides Android into three segments. Google will maintain control of the core OS segment, hopefully allowing for bigger, broader updates with less meddling from device manufacturers and carriers.
There are also new rules about user identifiers. Developers and advertisers use these to track users on and off their devices and serve up targeted ads. Restricting them is a privacy win for consumers, and probably a data win for Google, which controls most of the means of identifying users.
Another innovation in Android O comes in the notification center. Google designers and developers have paid particular attention to notifications and icons over the last few iterations of the OS. They rightly understand this is a major area of interest for consumers, and a primary area of interaction.
"Channels" is the internal term Google applies to new categories of notifications. In O, developers will group their notifications together into different types, letting users change how those notifications appear, or if they appear at all. Google's leadership says this is a feature requested by users and developers, and the benefits for both are clear. Users get more control over how their phone works — a key Android concept. Developers have more ways to communicate with users, and users will be less likely to simply uninstall apps that generate unwanted notifications. Instead, those notifications can simply be silenced.
That all comes with a catch: Once developers target Android version 26 (or Android O, to its friends), they must use notification channels. If they don't, their app's notifications will be dropped and simply not appear. This, of course, will have limited impact in the Android user community, given the low adoption rates for new Android OSes. But O will be a turning point for the platform and all developers.
Also during the session on notifications, the Googlers on stage took the unusual step of asking developers not to abuse the ability to change colors on notifications — a new feature in Android O. This warning was included in the developer documentation (and I noted it in my review of the pre-release of O), but it was strange to hear it on stage. It was also strange to hear Googlers say they were "giving a lot of trust" to developers, and promising to take away these privileges if they felt colorful notifications were being abused. It was odd, partly because of the paternalistic tone, but also because the vast majority of Android developers were not in the room or at I/O.
There are only two major tech companies with fully functional AR experiences: Microsoft through HoloLens and Google with Project Tango. Tango has been my perennial favorite at Google I/O in part because it is so radically different from other Google efforts, and also because it has felt ready for primetime since 2014 at least.
Tango had a big coming out this past year with the announcement of the first commercial phone to support it, the Lenovo Phab 2 Pro.
This year, we got word of an entirely new device with a thinner body and brighter screen. Lowe's has also agreed to let Google map 400 of its stores to provide in-store navigation via Tango devices that can take you directly to items on your shopping list.
From my conversations with the Project Tango team, the main focus right now is improving the overall experience and supporting the rollout of consumer applications for what was once a lab experiment. But we did get some hints. A session specifically on Tango showed off future efforts to identify discrete objects within the room, software that can map a room and then clear all the furniture from it, and an app to automatically generate 2D floorplans. One of the most exciting applications was an autonomous aerial drone using Tango technology to navigate obstacles at top speed.
The biggest VR announcement of Google I/O was, without a doubt, standalone VR devices. Google was short on details, but it's a marked departure from previous VR efforts. Google dipped its toe into the world of VR with the Cardboard, a $20 frame for a smartphone that provided a surprisingly immersive and low-cost VR experience. The company doubled down with Daydream, which used only high-end phones and a specialized headset, but still cost dramatically less than any other VR experience.
A standalone VR device effectively removes the hurdle of Google itself. It's a huge boon to Google and for the developers who want to try this new platform. With a standalone device, Google can offer VR to anyone willing to purchase such a device. You won't need a specialized Android phone; it will all be one package.
Daydream is very impressive in its own right, and I'm excited for a standalone device that will allow iPhone users and others to enjoy Daydream apps. Hopefully it will cost far, far less than the Vive and the Oculus Rift. But I worry about moving too far from the humble Cardboard. Part of what impressed me about it was its DIY efficiency and remarkably low barrier for entry. As the price of Google's VR experience goes up, the number of people who can enjoy it will go down.
In 2016, it wasn't clear what the AI revolution actually meant for Google. While machine learning was certainly at the forefront, there wasn't any tangible evidence of how this would work for Google. It could all have easily been dismissed as a fad, or just a nice way to encapsulate a sector of Google's progress that had flown under the radar.
This year, we finally saw what being "AI first" really means. It means Google in more places, it means developers being able to interact in new ways (and on new devices). And most importantly, it shows Google with far more focus than ever before. It seems very likely to me that this year specifically will mark a major turning point not just for Google, but for the entire industry.