Google’s prowess at AI and machine learning was clear on day one of its 2017 developer conference.
In his second Google I/O outing as CEO, Sundar Pichai confirmed the company’s strengths at AI and machine learning, including several new AI-first products such as Google Lens, and evolutions of Smart Reply and Google Assistant.
Pichai revealed a new beta of the next mobile operating system, Android O, adding that there are now around 2bn Android users worldwide.
He also said there are 500m active Google Photos users and that around 1.2bn photos and videos are uploaded to Google Photos every day.
Here’s what else we learned:
1. New beta of Android O revealed
The next Android operating system, known as Android O, has been unveiled. The new OS really is an attempt to make 64-bit-style computing flourish on Android devices.
It features what Google calls fluid experiences such as picture-in-picture tools, notification dots, autofill, smart text selection and a new dashboard console for developers.
A new PIP tool allows people to manage two tasks at the same time, providing a movable window for one of the apps that are open.
A key development is the introduction of TensorFlow to Android, the core of Google’s machine learning efforts, which will let developers push the boundaries of computing on small hardware. Google is also working on a new neural network API that will work with TensorFlow without compromising smartphone capabilities.
Google has also revealed a new security tool called Play Protect, which automatically scans devices to make sure they are safe to use. Google already scans 50bn apps every day.
The big problem Google faces is getting users to update their Android operating systems. While the company has around 2bn active Android devices in the world, unlike Apple’s iOS users, Android users are slow to update their software. Only 7.1pc of the active Android base are using the year-old Nougat or Android 7, while 31.2pc are running a two-year-old Android OS and 29.7pc are now using software that is four or more years old.
Will Android O be the game-changer Google is hoping for? We’ll have to wait and see.
2. Google Photos is growing faster
In less than two years, Google Photos has attracted 500m monthly users, which puts it up there with Facebook (1.8bn) and Instagram (600m) in terms of usage.
At Google I/O, the company launched three new features, including shared libraries where friends can see shots taken in real time, suggested sharing based on who is pictured, and Photo books, actual physical books featuring the best images in your gallery. Photo books are rolling out this week in the US on web, and on Android and iOS next week, starting at $9.99 for a 20-page softcover book and $19.99 for a hardcover.
This shines a lens on Google’s social media ambitions and, despite the enduring trauma that is Google Plus, this suggests that the company hasn’t given up the ghost on social, because, in effect, Google Photos is becoming a social network.
3. A Lens on Google’s AI and machine learning ambitions
‘Lens’ is actually the word because AI and machine learning are everything at Google right now.
Pichai said that the company’s image recognition systems are now better than humans. He revealed Google Lens, an application that is similar to speech search but uses algorithms to automatically identify and tag people in a photo. This may raise some red flags for privacy advocates but just shows you what technology is becoming capable of.
The CEO said that Lens, which can be best summed up as an in-photo search feature powered by computer vision, will soon appear in Google Photos and Google Assistant.
He revealed that AI is also making its way across Google’s servers and into its data centres. The key thing to glean from this is that AI is really about pushing the boundaries of computer science.
“When we started working on search, we wanted to do it at scale,” Pichai said. “That’s why we designed our data centres from the ground up and put a lot of effort into them. Now that we’re evolving for this machine learning and AI world, we’re building what we think of as AI-first data centres.”
4. Google Assistant is coming to Home and iPhone
Google Assistant vice-president of engineering, Scott Huffman, revealed that not only is Assistant becoming available in more languages (French, German, Brazilian, Portuguese and Japanese), it is also coming to the iPhone as a fully-fledged app.
Until now, Assistant was only available to iOS users through the Allo messaging app.
Google added new features to Assistant, including its integration into Google Home, to schedule appointments with your voice, make hands-free calls and access entertainment.
“Later this year, we’ll add visual responses from your Assistant on TVs with Chromecast,” Huffman said. “You’ll be able to see Assistant answers on the biggest screen in your house, whether you’re asking, ‘What’s on YouTube TV right now?’, or, ‘What’s on my calendar today?’”
5. Google is bringing Smart Reply to Gmail for Android and iOS users
More evidence of Google’s prowess at machine learning came in the news that the company is bringing its Smart Reply feature – currently in Gmail desktop, Wear and Allo – to Gmail on Android and iOS devices.
Smart Reply scans an incoming message and suggests at least three basic responses that the user can send, or tweak and send.
It will be available in English first, followed by Spanish and other languages.
6. Google is going to war with the hackers
After a week of WannaCry and the damage it has caused, the same kind of functionality that Google uses to keep spam and malware out of applications such as Gmail is about to become widespread across the Android ecosystem.
With more than 2bn Android devices globally to keep secure, Google has introduced Google Play Protect.
“Play Protect is built into every device with Google Play, is always updating, and automatically takes action to keep your data and device safe, so you don’t have to lift a finger,” explained Edward Cunningham, product manager for android security.
7. Kotlin promoted to first-class language for writing Android apps
Google said it is making Kotlin – a statically typed programming language for the Java virtual machine – a first-class language for writing Android apps.
Kotlin is 100pc interoperable with Java, which, up to now, was Google’s primary language for writing.
“We’re thrilled with the opportunities this opens up,” said Kotlin product manager Maxim Shafirov.
“For Android developers, Kotlin support is a chance to use a modern and powerful language, helping solve common headaches such as runtime exceptions and source code verbosity. Kotlin is easy to get started with and can be gradually introduced into existing projects, which means that your existing skills and technology investments are preserved.”
Starting immediately, Android Studio 3.0 will ship with Kotlin out of the box, without any extras or compatibility concerns for developers.
“It also means that moving forward, you can rest assured that both JetBrains and Google will be supporting Android development in Kotlin,” Shafirov said.
8. Job listings are coming to Google
Pichai announced that Google is launching a jobs search engine in the US, with roles ranging from entry-level service industry to high-end professional.
Utilising Google’s machine learning and AI platforms, the new search engine will seek to better understand how jobs are classified and related.
Pichai said that it will include a feature that collects and organises the millions of postings from all over the web to tailor them for jobseekers.
He said that 46pc of US employers face talent shortages and yet, there is a disconnect between employers and jobseekers.
“The challenge of connecting jobseekers to better information on job availability is like many search challenges we’ve solved in the past,” Pichai said.