This evening Google kicked off their annual developer conference Google I/O with the presentation of new technologies, products and software updates. Here are some of our personal highlights - we know there’s a bunch of other new stuff announced, but in this recap we focus on what’s most relevant to our work.
The keynote started with Google CEO Sundar Pichai mentioning AR wayfinding for Google Maps, a tool we fell in love with after we could test it a few weeks ago. Later on his colleague Sabrina Ellis, VP of Product Management, said that these AR walking directions are finally getting widely released on Pixel phones. What’s even better is what Aparna Chennapragada, VP Google Lens & AR, said right after: ‘With computer vision and AR, the camera in your hand is turning into a powerful tool’, and we couldn’t agree more. We have seen the value of combining AI (in computer vision) and AR in several of our use cases already, like our Telenet Cable classifier for example.
Google Lens, the tool that shows you information related to objects it identifies using visual input from your camera or pictures, has now been used more than a billion times. People use it to discover which flowers they’re looking at, where they can buy furniture they come across or to discover more about buildings they visit. In the future Google Lens will get even more powerful, said Chennapragada.
One of the new features will focus on language assistance: point the camera at a sign and it will read the text aloud to you. This isn’t exactly new, we know, but what is revolutionary is that Google made it fit in a mere 100KB file with support for 12 languages and it even works on really affordable phones. Your smartphone camera will also be integrated in Google Search, as it will become a new way of searching for specific keywords.
Google Assistant will also become way more powerful. With its new near instant offline speech recognition it can open and navigate your applications in the blink of an eye, allowing you to get even more done in less time, without stressing your phone or internet connection.
Next up they dived deeper into recent AI innovations. ‘We want to ensure that our AI models don't reinforce bias that exists in the world’, said Pichai. To reduce this bias, Google launches TCAV, a new model that shows what elements of the training data are important to reveal biases. Another new feature is the so-called ‘Federated Learning’, a new approach to machine learning, which doesn’t collect raw data from your devices in the cloud, but only transfers what it learned, so others can benefit from it as well. Sounds interesting, although our data scientists of course want to experience it first.
Talking about Android, Google focused mainly on privacy and digital wellbeing - currently the hottest topics in Silicon Valley. It’s interesting to see how Google is shifting its narrative towards the one of its biggest competitors, Apple, which made privacy its Unique Selling Proposition. Google will also bring the incognito mode we know and love from Chrome to Google Maps and YouTube. Building on the Digital Wellbeing announced last year, in the fall of this year, Google will introduce Focus Mode, to both Android P and Q, which will disable the apps of your choice you don't want distracting you when it's on.
All in all, the keynote at the beginning of Google I/O is becoming more and more targeted at a general audience. The announcements that are interesting for developers weren’t really that new - although they did mention some updates on what we can do with TensorFlow and BERT (a model for training Natural Language Processing models). The more in-depth talks on developer topics will come in the developer keynote later this evening, which we’ll recap tomorrow.