The Google I/O 2017 Keynote Live Blog

12:51PM EDT – Taking place today is Google’s annual I/O developer conference. Starting things off as always is the keynote, where we should receive updates on several Google technologies, initiaitves, and other Google-centric projects

12:51PM EDT – Android O will obviously be a big focus

12:52PM EDT – Android Wear and Android TV are also good bets

12:52PM EDT – I’d also be surprised if we don’t see something hardware related, though that could be just about anything

12:54PM EDT – So far Wi-Fi is behaving, and the weather is clear, if a bit chilly in the shadows. So we should be in for a good show

12:56PM EDT – This is the second year Google has held their event out at Mountain View’s Shortline Amphitheater; it still remains a rather unique event

12:58PM EDT – The keynote is scheduled to start at 10am, so we should be getting underway in a few minutes

01:01PM EDT – And here we go

01:03PM EDT – Google is opening things up with an animated trailer

01:06PM EDT – Now on stage is Sundar Pichai

01:06PM EDT – Over 7000 people here

01:06PM EDT – There are a further 400 remote events

01:07PM EDT – Over one billion active Google users per month

01:08PM EDT – Android has crossed 2 billion active devices just this week

01:09PM EDT – Keep in mind there’s only 7.5 billion people in the first place

01:09PM EDT – Sundar is now talking about Google’s heavy investment in machine learning

01:10PM EDT – Smart Reply is being rolled out to Gmail users today

01:10PM EDT – First subject: voice

01:11PM EDT – Google’s voice dictation word error rate is down to 4.9%, which is down from 6.1% just in the last 6 months

01:11PM EDT – Discussing how machine learning has allowed the Google Home device to use 2 microphones instead of 8, and still get good voice recognition accuracy

01:12PM EDT – Image recognition has similarly improved

01:12PM EDT – Google says their vision error rate is now lower than the human error rate

01:13PM EDT – New Google initiative being announced today: Google Lens

01:13PM EDT – Will ship first in Google Assistant and Photos

01:14PM EDT – Using image recognition to do smart things with information in photos. IDing flowers, inputting the SSID of a WiFi router, etc

01:15PM EDT – Google is rethinking their computational architecture and data centers. Google wants to build AI-first datacenters

01:15PM EDT – The core of this is Google’s Tensor Processing Units, which they started using last year

01:15PM EDT – Now recapping machine learning training versus inference

01:16PM EDT – Google’s TPU was optimized for inference. Training requires greater precision

01:16PM EDT – Google is announcing their second-generation TPU today, the Cloud TPU

01:16PM EDT – 180 TFLOPS per Cloud TPU chip. 4 chips per board

01:17PM EDT – Cloud TPUs are coming to the Google Compute Engine as of today

01:17PM EDT – These new TPUs are optimized for inference and training, implying that they operate at greater precision

01:18PM EDT – Google’s AI efforts will be coming together under the umbrella

01:18PM EDT – Research, tools, and applied AI

01:19PM EDT – Now discussing AutoML: using neural nets to design better neural nets

01:19PM EDT – “Learning to learn

01:20PM EDT – Now discussing applications for AI such as digital pathology

01:22PM EDT – Google also has projects for DNA sequencing, chemistry, and drawing assistance tools

01:23PM EDT – Now on to Google Assistant

01:23PM EDT – Recapping it, rolling a demo video

01:24PM EDT – Now on stage: Scott Huffman

01:24PM EDT – Google Assistant is now available on 100M devices

01:26PM EDT – Google wants to further improve the conversational abilities of Google Assistant

01:26PM EDT – Google is adding the ability to type to the assistant starting today

01:27PM EDT – Further down the line, Google will be rolling out Lens for Assistant, allowing it to talk about what it’s seeing

01:28PM EDT – Combining Word Lens image translation with more standard Assistant natural language queries

01:29PM EDT – Also new: Google Assistant is now available on the iPhone

01:30PM EDT – This is part of a larger effort to get it into more devices

01:30PM EDT – Google is rolling out an Assistant SDK to allow product developers to add Assistant to their products

01:31PM EDT – Starting this summer, the Assistant will be available in German, French, and several other languages

01:32PM EDT – Now on stage: Valerie Nygaard

01:32PM EDT – Valerie is here to talk about how developers have been adding features to Google Assistant

01:32PM EDT – Actions on Google

01:33PM EDT – Starting today, Actions on Google will be supporting transactions

01:33PM EDT – Account creation, ordering/purchasing products, and more

01:33PM EDT – Demoing ordering food via Google Assistant on a phone

01:34PM EDT – Payment by fingerprint (Google Payments, I’d assume)

01:35PM EDT – Now on stage: Rishi Chandra

01:35PM EDT – Rishi is here with news about Google Home

01:36PM EDT – Launching in Canada, Australia, France, Germany, and Japan this summer

01:36PM EDT – 4 new features to be rolled out over the coming months

01:36PM EDT – 1) Proactive assistance

01:37PM EDT – Google will be starting simple, and it will have multiple user support

01:37PM EDT – 2) Hands-free calling for Google Home

01:38PM EDT – You can call any landline or mobile number in the US or Canada for free

01:38PM EDT – (If this isn’t a prime example of how cheap VoIP to POTS has become, I don’t know what is)

01:39PM EDT – Will be rolling out over the next few months

01:39PM EDT – Also, you can dial out using your personal number

01:40PM EDT – 3) Entertainment. Spotify’s subscription and free services will be available on Home. SoundCloud and Deezer as well

01:40PM EDT – Bluetooth support is also coming

01:41PM EDT – 4) Today Google is announcing support for visual responses with Google Home

01:42PM EDT – Google Home can notify your phone (iOS or Android) and TVs via Chromecast

01:43PM EDT – Continuing to demo TV-related functionality

01:43PM EDT – Playing YouTube videos, showing the weather, etc

01:44PM EDT – Also integrates with Google’s recently launched YouTube TV service

01:45PM EDT – Now on stage: Anil Sabharwal

01:45PM EDT – The next subject is Google Photos

01:46PM EDT – Recapping Google Photos advancements in the last couple of years

01:46PM EDT – 1.2 billion photos and videos are being uploaded each day

01:46PM EDT – Google is launching 3 new features for Photos

01:47PM EDT – 1) Suggested sharing

01:48PM EDT – Google will suggest photos to share, and who to share them with

01:48PM EDT – Photos will have a new “sharing” tab

01:49PM EDT – Demoing the feature

01:49PM EDT – Photos will send photos via SMS or email if the recipient doesn’t have a Google Photos account

01:50PM EDT – 2) Shared Libraries

01:51PM EDT – Share part or all of your libraries with others

01:53PM EDT – Demoing how library sharing works and how it can be set to auto-save certain photos

01:55PM EDT – The new sharing features will be rolling out in the coming weeks

01:55PM EDT – 3) Photo Books

01:56PM EDT – It will now be possible to make photo books on the Photos application, with an emphasis on making it easy

01:56PM EDT – Demoing the feature now

01:57PM EDT – Google will be making the books; they offer softcover and hardcover books

01:57PM EDT – In the future, Google will be adding machine learning into the mix to suggest photo book compositions

01:58PM EDT – Photo books are available today on the website, and next week on the mobile apps

01:58PM EDT – Everyone at I/O will be receiving a free hardcover photo book

01:59PM EDT – Finally, Google Lens is being added to Google Photos

02:00PM EDT – IDing items in photos, etc

02:01PM EDT – Up next: YouTube news

02:02PM EDT – Now on stage: Susan Wojcicki

02:04PM EDT – YouTube has passed 1B hours/day viewed

02:05PM EDT – (So far this is more promotional than announcing anything new)

02:06PM EDT – Over 60% of watch time is now on mobile devices

02:07PM EDT – But living room viewing is the fastest growing segment

02:07PM EDT – Now on stage: Sarah Ali

02:08PM EDT – Google is adding support for 360 degree videos on the TV YouTube application

02:09PM EDT – Demoing controlling a 360 degree view using a TV remote

02:09PM EDT – (Looks like an NVIDIA Shield TV controller)

02:10PM EDT – Now on stage: Barbara Macdonald

02:11PM EDT – Barbara is here to discuss Google’s Super Chat feature, which actually launched earlier this year

02:11PM EDT – Demoing Super Chat

02:13PM EDT – Google is adding a new API feature to Super Chat to allow Super Chat to trigger device actions

02:15PM EDT – (This water balloon stunt is cringey)

02:17PM EDT – Now on stage: Dave Burke to talk about Android

02:18PM EDT – (2 Billion active devices; I have to wonder how many of them are up to date on security patches…)

02:18PM EDT – The 2B number is just phones and tablets, BTW

02:18PM EDT – Dave is currently recapping recent efforts. Android Wear, Android Auto, etc

02:20PM EDT – 82B apps and games installed via the Play Store in the last year

02:20PM EDT – Now focusing on Android O

02:20PM EDT – Release later this summer

02:21PM EDT – Dave is going to walk us through two themes that Google wants to focus on

02:21PM EDT – 1) Fluid Experiences

02:22PM EDT – Multitasking on Android O: Picture-in-picture on a phone

02:23PM EDT – Notification dots: a dot on the icon for an app to indicate that an app has a notification

02:24PM EDT – All fully automatic, without developers having to do extra

02:24PM EDT – Autofill with Google: autofill has been extended to apps

02:24PM EDT – Auto fill usernames, passwords, etc

02:25PM EDT – Copy & paste: smart text selection

02:25PM EDT – On device machine learning to intelligently automatically select text based on context

02:25PM EDT – Select a whole name at once, etc

02:26PM EDT – Announcing TensorFlowLite

02:27PM EDT – TFL will be leveraging a new API for neural network hardware

02:27PM EDT – The neural network will be made avaialble to O later this year

02:27PM EDT – 2) Vitals

02:28PM EDT – Vitals is focused on core functionality like security, performance, and battery life

02:28PM EDT – 3 foundational building blocks: security enhancements, OS optimizations, and developer tools

02:29PM EDT – Google is going to make their Google Play store security features more obvious

02:29PM EDT – Google Play Protect

02:29PM EDT – Letting users know the app has been scanned

02:29PM EDT – OS optimizations: faster boot time

02:30PM EDT – O will be adding limits to background execution

02:30PM EDT – Play Console Dashboards, to help developers see what problems users are happening

02:31PM EDT – Having

02:31PM EDT – One more thing

02:32PM EDT – Google is adding a new programming language to Android

02:32PM EDT – Kotlin

02:32PM EDT – Fully ART compatible, and interops with existing Android apps

02:33PM EDT – A ton more features they aren’t going into right now, such as Project Treble

02:33PM EDT – First Android O beta out today

02:33PM EDT – But wait, there’s more

02:34PM EDT – Now on stage: Sameer Samat

02:34PM EDT – Sameer is here to talk more about Android

02:35PM EDT – How do we get smartphones to more people and more of the world?

02:35PM EDT – Annoucning the successor to the Android One program, Android Go

02:36PM EDT – Android Go focuses on 3 things: optimized OS for low-end devices, smaller built-in apps, and a Play Store that highlights suitable apps (though all apps are accessible)

02:36PM EDT – For phones with 1GB or less of RAM

02:36PM EDT – Go devices can run with as little as 512MB

02:37PM EDT – Quick settings will include data usage information

02:37PM EDT – Chrome Data Saver will be on by default

02:38PM EDT – YouTube Go: a version of the app optimized for Android Go

02:38PM EDT – Save videos to watch them later. Peer-to-peer sharing of saved videos

02:40PM EDT – Improved keyboard tools for typing in other scripts

02:40PM EDT – “Building for Billions” best practices for Android Go

02:41PM EDT – Useful offline state, sub-10MB APK size, and better performance through GCM

02:41PM EDT – Starting with O, all 1GB or less devices will get the Go configuration

02:41PM EDT – And all devices will eventually have a Go configuration

02:42PM EDT – Now on stage: Clay Bavor to talk about AR and VR

02:43PM EDT – LG’s next flagship phone will support Daydream

02:43PM EDT – Samsung GS8/GS8+ will add Daydream support this summer

02:44PM EDT – Daydream will now support stand-alone headsets as well

02:44PM EDT – (Qualcomm has been pushing this idea particularly hard)

02:45PM EDT – WorldSense tracking technology

02:45PM EDT – Inside-out tracking from the device

02:45PM EDT – Google will be taking a platform approach for standalone Daydream devices

02:46PM EDT – Working with Qualcomm, of course. Also working with HTC and Lenovo on headsets

02:46PM EDT – Standalone devices later this year

02:46PM EDT – Update on Project Tango

02:47PM EDT – The second generation Tango phone will be the Asus Zenfone AR, which will go on sale later this summer

02:48PM EDT – Visual Positioning Service

02:48PM EDT – The phone finds its place by looking around and identifying visual markers

02:49PM EDT – CV meets positioning

02:49PM EDT – Adding a new AR mode to Google Expeditions for education

02:51PM EDT – And back to Sundar

02:53PM EDT – Sundar is sharing a developer story on TensorFlow

02:53PM EDT – Rolling a vido

02:56PM EDT – New initiative: Google for Jobs

02:56PM EDT – Job listing/matching service

02:57PM EDT – New search feature to help people find job postings

02:58PM EDT – Google’s worked with all of the major job listing services

02:58PM EDT – Filtering jobs by type, hours, etc

02:59PM EDT – Addressing jobs of every skill and experience level

02:59PM EDT – Rolling out in the US in the coming weeks

03:00PM EDT – Recap time

03:00PM EDT – Google’s shift from mobile-first to AI-first

03:00PM EDT – And that’s a wrap