Google Showcases AI Superpowers at I/O 2018: The Top 6 Announcements

Every year, tech giants and developers from around the world descend upon Mountain View for Google I/O, the company‘s annual developer conference. While the event covers a wide range of Google products and platforms, in recent years, one of the hottest topics has undoubtedly been artificial intelligence.

Google CEO Sundar Pichai kicked off the conference by discussing the company‘s vision of becoming an "AI-first" company, highlighting AI as the key driver behind many of its core products and newest innovations. From email to photos to digital assistants, AI is increasingly woven into the fabric of Google services that billions of people use every day.

At I/O 2018, Google awed the crowd with jaw-dropping demos that showcased just how far its AI capabilities have come, and hinted at where they might take us in the future. Let‘s dive into the top 6 most exciting AI announcements from the event.

1. Smart Compose for Gmail

Writing emails just got a whole lot easier with Smart Compose, a new Gmail feature powered by artificial intelligence. Building upon Smart Reply, which suggests brief responses to emails, Smart Compose takes things to the next level by auto-suggesting complete sentences as you type.

To generate these suggestions, Smart Compose leverages machine learning to understand the context and style of your email based on the text you‘ve written so far. As you‘re typing, it presents suggestions for how to finish your sentence or thought. Simply hit Tab to accept the suggestion and keep typing away.

Smart Compose has the potential to drastically speed up email writing, especially routine messages where you find yourself typing the same phrases over and over. It adapts to your personal writing style and can even pick up on your unique greetings and sign-offs. The more you use it, the smarter it gets. And of course, you always have the ability to ignore the AI‘s suggestions and express yourself in your own words.

By taking care of the mundane aspects of composing messages, Smart Compose frees you to focus on crafting thoughtful, personalized emails in less time. It‘s a stellar example of how AI can boost everyday productivity in a way that still maintains a human touch.

2. ML Kit

Many of the coolest AI-powered features in apps rely on sophisticated machine learning models for tasks like recognizing objects in photos, detecting faces, or converting speech to text. However, developing this kind of functionality from scratch can be time-consuming and challenging, especially for smaller app devs or companies without AI expertise.

Enter ML Kit, a shiny new SDK from Google that enables Android and iOS developers to easily integrate pre-trained machine learning models into their apps. ML Kit contains a set of ready-to-use APIs for common mobile use cases:

• Text recognition – Detect and extract text from images
• Face detection – Detect faces and facial features in images
• Barcode scanning – Scan and parse data from most barcode formats
• Image labeling – Identify objects, locations, activities, and more in images
• Landmark recognition – Identify popular landmarks in images

For developers who want more customization, ML Kit also supports TensorFlow Lite models, allowing you to compress your own models and run them on-device. This means you can build intelligent features unique to your app while still benefiting from Google‘s machine learning tools and infrastructure.

With ML Kit, Google aims to democratize AI development and enable a wider range of apps to take advantage of machine learning. By lowering the barrier to entry, it could spur a wave of creative AI-powered features and experiences on mobile.

3. Google Duplex

The clear showstopper of the I/O keynote had to be Google Duplex, a jaw-dropping demo of just how convincingly AI can mimic human conversation. Duplex is an extension of Google Assistant that can make phone calls and carry out "real world" tasks on your behalf, like booking a hair appointment or making restaurant reservations.

What makes Duplex so remarkable is how astonishingly human-like it sounds, complete with the "ums" and "ahs" of natural speech. If you heard the audio clips of its phone calls out of context, you‘d easily believe you were listening to two humans conversing.

At its core, Duplex leverages several AI technologies including natural language processing, deep learning, and text-to-speech. It grasps complex sentences, handles interruptions, and responds with the natural cadence of human speech.

Google was careful to point out that Duplex isn‘t intended to replace human-to-human conversation, but rather to carry out very specific tasks in situations where a phone call is required. There are still many open questions around how Duplex will work in the real world and address valid concerns around privacy and disclosure to the humans on the other end.

Nonetheless, Duplex is an undeniably groundbreaking leap forward for conversational AI. It opens the door for digital assistants and chatbots to sound more natural than ever before. Whether you find it astonishing or a bit eerie, Duplex provides an exciting glimpse into a future where AI is increasingly woven into the fabric of everyday life to help with mundane tasks.

4. Suggested Actions for Google Photos

Google Photos is getting smarter by the day with the help of AI. Suggested Actions is a new feature rolling out to Photos that uses machine learning to offer intelligent, contextual actions you might want to take on a photo.

For instance, let‘s say you snap a picture of your child blowing out the candles at her birthday party. Photos can recognize the low lighting in the image and automatically offer to brighten it up. Or if the photo contains several people, it can suggest sending the picture to those individuals.

In another example, Photos might recognize a document in an image and offer to convert it to a PDF for easy sharing or printing. If you‘re a photo hoarder like me, this could be a game changer for finally organizing that dusty digital shoebox of receipts and notes!

In addition to Suggested Actions, Google also announced several other AI-driven updates to make Photos more powerful. The app will now colorize old black-and-white photos with impressive accuracy. It will convert photos of documents to PDF and enable you to search them by text. And a "Favorite" feature uses AI to surface your best shots.

With over 1 billion users and 5 billion picture uploads per day, Google Photos offers an ideal playground to flex Google‘s computer vision and machine learning muscles. These new features aim to take the hassle out of editing, organizing, and sharing an ever-growing library of precious memories.

5. Google Maps

Google Maps is a lifeline for navigating the world around us, and now it‘s getting supercharged with AI smarts. At I/O, Google previewed a series of upcoming enhancements to Maps powered by machine learning and augmented reality.

Soon when you‘re exploring an area, Maps will surface contextual suggestions for events or activities you might be interested in, like a concert or new restaurant. It stitches together information from the places you‘ve been, events you‘ve searched for, and cuisines you‘ve sampled to offer helpful recommendations tailored just for you.

If you‘re meeting up with a friend, Maps can show you their real-time location and estimated time of arrival, so you know if you have time to grab a coffee before they show up. For those of us with a terrible sense of direction, a new AR walking navigation mode uses your phone‘s camera and Google‘s Street View imagery to visually guide you to your destination with floating arrows overlaid on the real world.

Under the hood, Maps is harnessing AI and the power of the cloud to make sense of the billions of data points feeding it every day – traffic patterns, business info, user photos, and more. By weaving all of this intelligence into a cohesive experience, Google aims to transform Maps into an indispensable sidekick for navigating not just geographic space, but your social life as well.

6. Google Lens

Google Lens, the company‘s visual search and analysis tool, is gaining a slew of new AI-powered features to help make sense of the real world. Point your phone‘s camera at an unfamiliar sign or document in a foreign language, and Lens will detect and translate the text instantly. Snap a picture of a business card, and Lens can parse the key details and export them as a new contact.

One of the coolest additions to Lens is Style Match, which lets you shop for fashion and home decor products simply by pointing your camera at them. Using computer vision and machine learning, Lens analyzes the item in question – say, a sleek pendant lamp or eye-catching sundress – and presents similar products available for purchase online. It‘s like Shazam for the real world.

Lens is also expanding its integration with other Google services. When you‘re looking at a web page in Chrome or an email in Gmail, you‘ll be able to invoke Lens to analyze an image and pull up relevant information. And coming soon, searching your Google Photos library with Lens will be just a tap away.

With every new release, Lens gets closer to its original vision of being a visual browser for the real world – an intelligent assistant that can understand what you‘re looking at and take helpful actions. As Google combines advancements in computer vision, natural language understanding, and knowledge mining, Lens is steadily becoming an indispensable tool for navigating the blurring lines between the digital and physical worlds.

The AI-First Future

Google‘s announcements at I/O 2018 made one thing abundantly clear: artificial intelligence is no longer a nice-to-have. It is the driving force behind virtually all of the company‘s core products and the key to unlocking more intuitive, immersive, and intelligent user experiences.

In the AI-first era Google is charting out, our computing interactions will become increasingly natural and conversational. Our apps and services will grow ever-more capable of understanding context and catering to our needs. The digital and physical worlds will continue to converge as AI helps us navigate between them.

Of course, Google isn‘t the only tech giant investing heavily in AI, and many of the advances previewed at I/O are still years away from being fully baked into products. But if one thing is certain, it‘s that Google‘s AI ambitions know no bounds. As it weaves artificial intelligence deeper into its sprawling ecosystem, it‘s safe to say our technological future will be anything but artificial.