We've barely scratched the surface with AI

WHEN artificial intelligence (AI) is brought up today, a common image that comes to mind for many people is a chatbot or a translator. AI is all that, but more.

Sometimes called machine intelligence, AI is a discipline that has been around for decades but is gaining momentum thanks to the convergence of three major forces - increased computing power in the cloud, powerful algorithms that run on deep neutral networks and access to massive amounts of data.

Amid this explosion of data, we are still constrained by our human capacity to process and make sense of it. The question is, how can we use all we have in terms of computational power to solve this fundamental constraint, and to make better sense of the world? That is the essence of what AI is. It is not about having AI that beats humans in games; it is about helping everyone achieve more - humans and machines working together to make the world a better place.

Yet, with the speed at which technology is advancing, we are only barely scratching the surface when it comes to the possibilities of tomorrow.

At the heart of all AI applications is data - but instead of humans taking on the onerous task of understanding the data, we train AI tools with algorithms to learn from data and provide "machine" intelligence.

Today, smart search powered by such algorithms enable enterprises to analyse, interpret, transform and enrich data that would otherwise be unconnected and hard to make sense of. For example, Microsoft Graph allows businesses to use unique data from their organisation to drive workplace transformation.

For example, My Analytics in Office 365 is like a fitness tracker for your workday, showing how you are spending your time, informed by the Microsoft Graph and powered by AI.

At Microsoft, we believe that people around the world can benefit from AI - but only if AI technologies are available for them.

While developing a chatbot was once considered complex and costly for organisations, today there are many "toolkits" available that make building a chatbot akin to putting together a Lego set.

With Microsoft Cognitive Services, developers can build apps that recognise gestures, translate text into multiple languages, deconstruct video for quicker search, provide editing and real-time captioning, and even customise data to recognise images in categories that are most important to customers without developing it from scratch.

As we infuse intelligence into everything, we aim to take the same capabilities and make it available as a set of cognitive services to every developer and democratise it so that every one can use the same building blocks that we use to build Office 365, Cortana and Dynamics 365.

The world that we are living in is fast becoming one of an intelligent cloud, and the intelligent edge.

The common perception of AI is that it is centred on a big brain on the Internet, where all questions are answered. This may be true, say, of early versions of virtual assistants, but increasingly, the industry is seeing the need for AI that is embedded in devices which users rely on even when they are offline - in other words, AI that is "local" or on the edge of a network that may not be always connected.

To do this, devices that use specially-designed hardware accelerated for AI are now able to run advanced deep learning algorithms directly and on the edge of a network - without the heavy computational power or connectivity to a central server for analysis, as required in the past.

Chipmaker Qualcomm's Snapdragon Neural Processing Engine hardware and its Vision Intelligence Platform, for example, offer blueprints for devices that can collect data and learn from it while being connected to an efficient network designed for AI, such as Microsoft's Azure IoT Edge.

By optimising the hardware as well as providing an efficient connection at the edge, this new solution is changing the way that AI is deployed throughout an organisation, by rolling out more intelligent sensors, with various levels of AI capability, based on this new model.

Indeed, these new applications of AI are already changing lives in areas where the Internet is not widely available.

AI-powered insights

In India, doctors from Focus Health have used smart retinal imaging screening devices to check patients living in remote areas for common eye problems that could lead to blindness. With AI built into the scanners and connected to Microsoft's Azure IoT Edge network, operators can draw on AI-powered insights to determine if a person has eye problems. This is done by analysing the images taken of each person's eyes.

Including India, Focus Health has deployed its 3Nethra portable scanner in more than 20 countries, making a difference to more than two million patients. By detecting symptoms early, patients can seek treatment to avoid the loss of vision.

Such developments are accelerating AI adoption, by democratising the technology through easy-to-deploy solutions. They are also highly customisable, so enterprises can fulfil their specific needs.

Another example of this is Ascendas-Singbridge, which rolled out a mobile app called Ascendas-Singbridge App, or ASAP. This app offers an easy and swift interface for tenants to report faults in their buildings, as well as to check for the latest retail promotions in the building, shuttle bus information and even interactive building directories.

The company also built the Ascendas-Singbridge operations centre, known as AOC, which serves as a "brain" that senses and monitors the operations in the buildings that the company manages, enabling real-time status updates on essentials such as electricity, air-conditioning, lifts and water supply. If there are any disruptions, a team is quickly dispatched to fix the problem.

Working with Microsoft's Smart Building Solutions and Azure Machine Learning solutions, the company uses data analytics to predict when equipment might fail, backed by a proactive maintenance programme. Video analytics is also enabling the company's current pool of security guards to work smarter and more efficiently, as well as to manage a smart carpark system that can predict surges in usage and open more hourly parking lots in anticipation of the higher number of visitors' cars.

These advances underline the tremendous momentum that AI has gathered in the past two years. Instead of being a simple answer engine, AI is now clearly ready to tackle more complex issues, starting with understanding all the data that has been collected. From this, AI will grow to be everywhere in an enterprise, from a camera that is smart enough to understand what it is "seeing" in real time, to portable devices searching for early signs of illness so that people may seek life-saving treatment.

With the impact that AI is having, it is no wonder many industry experts view 2018 as the year of AI. But we are only scratching the surface of what AI can help us accomplish.

AI will enter the mainstream, transforming lives in a meaningful way that was not thought possible just a few years ago; and it is only getting started.