Google I/O 2018 keynote
Disclaimer: The following content is based on personal recollections and interpretations and may therefore be incomplete or faulty. Attendance was funded by me as a private individual so views and opinions do not reflect those of my employer. Content licensed under Creative Commons (CC BY) Casper Bang.
I have returned to Google's backyard, the outdoor Shoreline Amphitheatre venue, for the 2018 Google I/O festival. As a senior Android developer, there is simply no better way to spend time and money than getting to know the latest and greatest technology while also networking with Googlers and peers.
Like the previous years, the first day is split into 3 core activities; Google keynote by CEO Sundar Pichai, Developer Keynote by various program managers and finally the many smaller discrete technical sessions which will also dominate the following 2 days. This post will focus on the Google Keynote - Developer Keynote, talks etc. will follow in another post.
Google Keynote
Unfortunately once again I was not one of the 7000 participants with a place in the shade, so on with the sunscreen, cap and sunglasses. There were no real major revelations in this years Keynote (unlike last years commitment to the Kotlin language) and everything was more about refinements and iterations on existing stuff.
Google Assistant
Google used machine learning to develop a more natural speech synthesization and added new natural sounding voices for the Google Assistant thanks to their WaveNet technology, which is now also able to accept multiple commands at once as well as carry out longer contextual conversations - without you having to invoke it with "Hey Google" again (the mike remains actively listening for 8 seconds after the latest feedback was given). One of the more dystopian demos was when we saw their Duplex technology do truly autonomous work like booking a haircut appointment and making a restaurant reservation.
At this moment, I find myself sitting at Google I/O 2019 wondering if Google Assistent on my phone will be screening calls, ensuring telemarketers and mother in law (just kidding Jacqueline) are filtered before the actual call reaches me. :)
Update: Since writing this article, I've spend a little time investigating and it's deeply impressive what's going on here once you think more about it. Machine Learning, deep learning and neural nets aside - this is starting to truly be something we can call AI.
Smart Displays
Staying within the Google Assistant theme, leads straight into one of the few actual hardware products which were - not exactly launched yet, but more teased about - namely Smart Displays. It is suppose to add visual output devices for the assistant experience like we have audio devices in the Google Home products. It remains to be seen just however, just how this is different than placing a tablet on a tablestand and it doesn't seem particular original (Amazon's Echo ecosystem already sells similar devices)
GMail Predictive Text (Smart Compose)
GMail is gaining yet another element of intelligence by the addition of Smart Compose - a feature which will offer to autocomplete your sentences as you respond to email within GMail. It remains to be seen whether this is an assistance or a pain in the ass, but it's an interesting idea and I have no doubt that people who write a lot of similar responses will benefit from this. I don't recall it mentioned by Sundar but it strikes me from his example that the Smart Compose is adaptive by nature, as in learning and adhering to the users style of language. Needless to say... Machine Learning appears to be at work here again. Smart Compose is rolling out in the coming weeks.
Android P
Curious early adopters have already been able to play around with Android P Preview - but today Google released a new Android P beta. Before diving into Android P news itself, I think it's worth mentioning that last years Project Treble (Google's AOSP restructuring and modulization project with OEM's and SoC vendors) already has a positive effect - betas as available for more than just Google's own Pixel reference devices, now supporting models from Essential, Nokia, Oppo, Sony, Vivo, and Xiaomi. Awesome to see!
As to Android P itself - they appear to now do away with the bottom navigation bar and the 3 buttons we've seen since it had 4 buttons. :) It is now much more gesture based, not dissimilar to what has been done at a certain fruit-named company in Cupertino. This is bound to happen, as smartphones are finally on the way of becoming just one giant screen and nothing else.
Machine Learning takes the stage again, in adaptive brightness as well as battery optimization being handled by on-device neural nets - it remains to be seen how well this works, but the more smarts that works out of the box, the better. I'm going to guess its based on the new Neural Networks API which was introduced last year in Android 8.1.
Google News
Google have had the Google Newsstand app for a while, but it is now being replaced by the Google News app for iOS and Android. It will roll out to 127 countries within a week according to Google. It should function as a smart news aggregater from 1000 different sources, utilizing machine learning to prioritize and provide alternative angles on news stories. Interestingly, users won't be able to customize which sources those are - Google will display the same information from the same sources to everybody. At first glance, this seems out of trend of adapting to the users preferences - until you think about it; the very thing Google News wants to address is non-skewed information where bias is handled by AI rather than the subjective opinions of the user.
Update: I was not able to see Google News on the Play Store (but I can see that Google Newstand app is no longer there) so I side-loaded it through ApkMirror here. Even though I am running with a Danish locale, it's clearly already translated and ready to serve a Danish audience as well - as evident below.
I have to say, the app feels super nice and I have no doubt it's going to be my go-to app for getting up to speed on just about anything.
Google Lens coming to Google Photos - probably
Google Lens - a smart assistant feature for images - actually launched at last years Google I/O as a feature Google Photo's, but this appeared to have been mostly a paper launch because we certainly did not see much to it at all during the year. Now it's also moving into the camera app. It's apparently going to launch on "supported devices" which I interpret as Stock Android (AOSP and Android One) devices.
Google Maps
Google Maps is adding a navigation mode for pedestrians; with the map being overlayed on a live camera stream with augmented markers. Something we have all been lacking forever when trying to orient ourselves and our phone with actual map directions.
Waymo
Google self-driving car project Waymo was on the stage for the first time at a Google I/O keynote and presented some of their current test setups and plans, but ultimately there weren't much news here - but the car designs have definately evolved positively and yes that's a self-driving truck you see on the picture.
Swag
Without any major gadget launches, swag was relatively moderate this time around - everyone got a Google Home Mini. If you cared to do the Scavenger hunt as well, you also got an AndroidThings development starter kit complete with microprocessor board, touchscreen, camera module and sensor/notification board.
All attendants also got a free eBook "Kotlin in Action" as well as free registration of an .app domain for the first year.
Impression
All in all, perhaps not quite as impressive as last year. Its understandable however, that no company can come out with revolutionizing new gizmo's and gadgets every year. Truth be told, the most interesting stuff to a developer is not in the Google Keynote anyway - it's in the all the following engineering sessions void of management and marketing speak and Google I/O is after all first and foremost a developer conference.
Since this writeup, I've added a more developer centric article. Check it out here.
Senior Engineering Manager at A.P. Moller - Maersk
6 年I think this year's announcements are still very impressive, this year was more about, to what extend can you use last year's revolutionary announcements in production. And i found some of it really amazing on how easy but yet efficient everything is becoming. The voice assistant, tensorflow, kotlin, ML kit, Android things. Can you believe those are between 1 and 2 years old products??? It's crazy!!
Founder & Director | AI Venture Network | A Not-For-Profit AI/ML Venture Acceleration Community | Next AIVN AI StartUp & Innovation Showcase for Founders and Funders March 20 | +1 602-345-0811 | Scottsdale AZ
6 年#worthless