State of Mobile November 18, 2016

Machine learning in apps

Hannes Van de Velde

Director of Product Design

In our annual report 'The State of Mobile' we have identified five very concrete evolutions you should understand to stay on top of the digital game in 2017. The second is Machine Learning in apps. Machine learning and Artificial Intelligence have been around for as long as 60 years, and have ever since been ‘right around the corner’ as the next big thing in computing. But since a couple of years, it looks like AI is finally able to live up to its esteemed potential.

What is happening? 

Over the course of the last years, some impressive innovations are putting machine learning back on the radar. And if we look at the big players, we have every reason to believe that this trend is ready for a major breakthrough.

The recent advances in machine learning are mainly caused by three crucial factors becoming available at the same time:

  • breakthroughs in deep learning techniques and neural networks, dramatically reducing the error rates of machine learning algorithms;
  • increasingly more connected devices generating huge amounts of (observed) data that can serve as training data for the learning algorithms;
  • faster computer processors and hardware (CPU, GPU and memory alike) that are able to process all this data.

Download The State of Mobile

Get insights into today’s mobile behaviour and technologies to watch in the years to come

Download the report
Download the report

How is the trend evolving? 

Machine Learning and Artificial Intelligence will impact the way we use technology in a couple ways and paces. In a first phase, machine learning will drastically change the way we look at digital interfaces. A second and more profound way of AI-disruption will come in the form of entirely new use-cases that will not only change the way we look at interfaces or technology, but at our entire lives.

Source: Britisch Columbia

Improved interfaces

The most straightforward way we’re already seeing machine learning impacting the products we’re using on a daily basis, is that their graphical interfaces are becoming more intelligent. By leveraging machine intelligence, we can anticipate on what a user might want to do:

  • Predictive keyboards and e-mail apps are using natural language processing to predict your next words, so you don’t have to type them yourself;
  • Photo apps are using intelligent image recognition to automatically tag, sort and group your pictures based on people or objects that appear on them or locations they were taken;
  • Facebook and other feed-based interfaces are leveraging machine learning to adapt and tailor your experience based on every interaction you have.

These early and very focused examples of Machine Learning are mainly aimed at helping people save time or creating more relevant experiences. But even these small intelligent interventions can have a huge impact on how we experience these products. By focusing on tasks that are very hard for humans but fairly easy for computers – typically highly repetitive tasks like scanning through lots of data and making sense of it – just a hint of artificial intelligence can add a great amount of delight and added value.

New interfaces

The rapid advances in natural language processing – making computers able to understand and interpret human languages – will eventually make graphical user interfaces obsolete for some use-cases. The graphical user interface has always been a necessary layer between humans and computers, in order for them to understand each other. But when computers are able to understand our ‘native interface’, human language, we will be able to interact with them in easier and more natural ways. In the coming years we’ll see conversational interface, mostly in the form of chatbots, replacing some of the graphical interfaces we’ve been interacting with in the past years. The biggest advantage of these conversational interfaces is that we can express ourselves in a very natural way, without being limited by the constraints or flows a graphical interface imposes.

Source: The Conversation

New use-cases

In a second parallel but slightly slower track, Machine Learning will herald a series of previously impracticable or unthinkable use-cases. Self-driving vehicles are a great example. Where it was previously unimaginable that a computer would be able to autonomously drive a car in a safe way, machine learning and artificial intelligence are making this a reality today. But the impact machine learning will have on technology and mankind as a whole will go way beyond that. Intelligent machines will be able to diagnose diseases better than any human doctor ever could.

The successful machine learning use-cases in the years to come will be based on tasks that are hard for humans, but easy for machines.

Project planning of large construction sites could be done by intelligent systems that are able to take into account more parameters and variables than a human project manager could. AI-powered personalised banking dashboards and assistants could bring entirely new insights and products to personal finance.

Do's & Don'ts

Download The State of Mobile for the do’s and dont’s in Machine Learning and insights into other technologies that will shape your business in the years to come.

Download The State of Mobile

Get insights into today’s mobile behaviour and technologies to watch in the years to come

Download the report
Download the report