Prediction is becoming cheaper, faster, more automated and more ubiquitous in our lives, and we have artificial intelligence to thank for that.

AI is essentially a predictive technology. No matter what its algorithmic underpinnings, its core function is to make sophisticated inferences about what’s likely to happen based on myriad variables that have been distilled both from historical and real-time data. When it’s embedded in every device and refined continuously with fresh data, AI becomes a ubiquitous resource helping us all to anticipate what’s coming and do what’s necessary to keep our lives running smoothly.

As McKinsey Analytics stated in this recent study, AI’s transformative power is in its ability to drive down the cost of embedding predictive capabilities in everything we do. “As the cost of prediction continues to drop, we’ll use more of it for traditional prediction problems… because we can predict faster, cheaper and better. At the same time, we’ll start using prediction to solve problems that we haven’t historically thought of as prediction problems.”

That study was ringing in my mind this week while absorbing the news from two major AI-focused industry conferences: Microsoft Build and Google I/O. It occurred to me that the news that achieved the greatest traction in headlines was all around the embedding of predictive intelligence in smart devices, sensors and infrastructure.

Essentially, this week’s banner headline was the emergence of anticipatory assistance as a standard feature of the smart devices that are starting to rule our lives. In other words, the trend is now for all devices to perform dynamic look-aheads on fast-changing environmental conditions and then use those anticipatory insights to drive appropriate actions.

We can see that in the following aspects of this week’s key announcements, most of which pertained to new features in mobile devices:

• Anticipatory devices: Devices will anticipate what users are trying to do and help them take those actions faster or in an automated fashion. Android P, the next major version of Google’s mobile operating system, will use embedded AI to anticipate what settings for battery life, screen brightness and applications will make best optimal use of device resources while helping users maintain acceptable experience. In addition, Android P can learn a user’s behavior to anticipate what’s the best app, song or action to active on the phone. Likewise, Microsoft Azure Cognitive Services will help developers create intelligent apps that anticipate and suggest text and other input modalities to help disabled people interact more effectively with devices. Embedded predictive AI will enable devices that run Microsoft’s Azure IoT Edge to take the right actions autonomously without cloud connectivity.
• Anticipatory conversations: Conversations will anticipate what users are trying to say and help them communicate their intent more rapidly and effectively. Using embedded AI, Google Assistant, its digital assistant, will automatically sense when users want to carry on extended conversations, thereby sparing the user from having to repeat the keep-alive phrase “Hey Google,” and will also proactively help users recall the thread of the conversation. Google Duplex, the new natural-language agent in Assistant, will anticipate a user intent within conversations, based on its ability to automatically dissect longer remarks, complex sentences and rapid-fire speech. Duplex can carry out sophisticated conversations and tasks fully autonomously, continuously anticipating which operations it can or cannot complete without user assistance, and reaching out to users in the latter scenarios. And users of Microsoft’s AI-powered mobile, embedded, IoT and mixed-reality apps will be able to tap into the anticipatory experiences that developers build into conversational device user interfaces, leveraging the 100-plus new features being shipped with Microsoft’s Bot Framework.
• Anticipatory contexts: Apps will anticipate the dynamically shifting contexts and intents of users’ engagement. Using embedded AI, Google’s Maps app will generate anticipatory AI-personalized navigation recommendations in real time and on users’ devices, superimposing geocontextualized display of place names, street names and directions in the smartphone camera view. Google’s Gmail app will use AI to help compose emails, suggesting sentences as a user types. Using embedded AI and graph technologies, Microsoft’s mobile, IoT and enterprise products will be able to adjust the UIs, actions, and other aspects of the user’s “multidevice/multisense” experience dynamically. And using the in-preview Microsoft IntelliCode, Visual Studio programmers will leverage an AI-powered autocomplete feature to receive anticipatory code suggestions inline to their keyboarding of a variables, function names and other code. The feature’s suggestions will come from AI that has been trained on a large GitHub code repository, offering context-aware relevance ranking of code suggestions.

To the extent that AI applications drive their predictions from deep historical data and continuously optimized statistical models, users will treat them as a natural adjunct of our organic intuitions. Going forward, the “A” in AI will, in many people’s minds, stand for anything but “artificial.”

If you need a new “A” word to plug into this acronym, anticipatory would make great sense. But so would assistive, augmented, anthropomorphic, accelerated, adaptive and automated. All of these describe the new world of AI-driven mobile applications that are continuously predictive and experiences that are, as a result, consistently optimized.

Here are Wikibon Chief Analyst Dave Vellante and me discussing these trends on theCUBE, SiliconANGLE Media’s video studio, late last year:

Image: geralt/Pixabay