Google IO 2017

Google IO 2017

Google IO is Google’s annual developer conference held in San Francisco. This year I attended Google IO Extended which happens all around the world at the same time as the main IO event, it’s designed for people who can’t make it to the main event but want to know the latest stuff.

There was one main theme this year from Google, and it’s summed up in this phrase:

“Mobile first to AI first”

In every area that Google spoke about (from new processing hardware, home automation to Android devices) everything had been improved by AI!

Another nice fact they mentioned was, Android now runs on over 2 billion devices and 82 billion apps were installed last year.

Below are some of the big headlines!

Google Lens

A new app designed for your phone, point it at something, be it a flower, restaurant sign or a WIFI label and it will understand it, identify the flower, show the menu for the restaurant or automatically join the WIFI! It can also translate languages on signs.

They also showed a cool demo where the AI could detect obstructions (a wire fence) and remove it from the picture). This is a huge leap in computer vision.

Google Home

Google home seems to do a lot more than I realise, for instance, it can recognition up to 6 different people in a household and customise the experience for each one. Now, Google is adding phone calling to Google Home for free. Only available in the US currently, you can just ask Google Home to phone your mum for instance and will recognise who you are, and find your mum in your contacts. If your partner does the same thing, it will phone their mum, not yours.

Another new feature is visual responses, which is super cool. You can ask Google Home something, say “what’s my calendar look like today”, and Google will display it on a Smart TV, Chromecast or Google connected device. I really think this will become super useful. You could ask Google Home, how long it will take to get to somewhere, then tell it to send directions to your phone.

They also introduced something called Proactive Assistance, the idea is that Google Home will detect things that may be important to you and let you know about them via a visual light on the device, for example, if traffic is really bad and you have a meeting coming up soon.

Google home now integrates with over 70 smart home manufactures

Virtual Reality

Google already make a VR framework (Daydream) and a headset to fit onto your phone, this year Google announced 2 stand alone (no phone, pc etc needed) VR headset coming out this year and have partnered with HTC (who make the HTC Vive VR headset) and Lenovo who make their project Tango tablet (3D mapping / AR). What’s very interesting here is that they are bringing out their own indoor tracking solution that does not need external sensors. They call it VPS (visual positioning system) which I believe could be an advanced version of SLAM.

They also announced that the new Samsung S8 will support the normal Daydream VR headset, which I found odd as Samsung are in partnership with Oculus (Facebook, direct rivals with Vive) and already have the GearVR.

Augmented Reality

Google announced another Tango handset (it’s like a Microsoft Kinect embed into an android tablet) and announced Expedition, which brings AR to the classroom. Kids will be able to place 3D augmented objects within the classroom, for example see how volcanoes explode.

Suggested Sharing

Suggested sharing is a new feature for Google Photos that uses AI to detect well-taken pictures, and who is in them. It then suggests / reminds you share that picture with the people in it. It forms an online collection of all the images, so you finally get to see images with you actually in them (if someone else took them). There is also an automatic mode, for example if you always want to share pictures of your kids with your partner. Feels a little scary to me.

Cloud TPU’s

So, anyone in computing will know what a CPU (central processing unit) and a GPU (Graphics Processing Unit) is. Google likes to do their own thing and last year announced the TPU (Tensor processing units) which are designed to be very quick at machine learning processes. Google are now calling them Cloud TPU’s and each one can do 180 teraflops.

Android O

There were a few new features mentioned in the keynote but nothing I found too exciting. They mentioned picture in picture, and notification dots, both of which iOS already have. They mentioned Android Studio 3 and supporting kotlin as a first class language, again, I guess it’s their answer to Swift for iOS. There was the usual focus on battery useage, security (Google Play Protect) and making apps run boot faster. They say they have seen 2x improvements on apps running. Google has also improved Copy and Paste features so that it automatically recognises address, company names, phone number etc which in all honesty I thought it already did.

iOS Support

Throughout the presentation, whatever new stuff they demo’d they kept making a point that it’s also supported on iOS, not just Android (Google Assistant, Google Photos, Daydream etc) which I personally thought was cool.

Lastly and probably the one that made me laugh the most!


Youtube for TV and consoles will now support 360 video including live events, Youtube viewing on TV has gone up by 60%. However, the big news is Super Chat and Trigger Actions.  

Super Chat allows you to pay for your comment (to a Live Youtuber) to be noticed, so if you really want to ask that question, you can pay for it. Not too bad, I guess. But Trigger Actions allow you to pay to trigger something in the live video, so throwing a water bomb at the presenter or turning the lights off in their house. I can see this going down hill pretty fast.