Machine learning: Making smartphones smarter

Machine learning is an application of artificial intelligence which enables automated improvements without explicit programming. It has accompanied every evolution of computing, since the first room-based computers built in the 1950s. 

It is likely to play an increasingly important – albeit often invisible – role in the evolution of smartphone application and handset design over the coming years. 

What is machine learning?

Machine learning uses a range of inputs to customise applications to the user. For example it could help train a voice recognition app, by helping it recognise the user’s voice. It can also use an array of other data points, including location history, app usage history, common words used when typing, common numbers called and photographs taken. 

There are already multiple smartphone applications that incorporate machine learning, but most users may be unaware that their smartphone usage is being enhanced by machine learning. 

Current levels of adoption and usage

Our research polled awareness and usage for a range of applications which use machine learning. Overall awareness was high: over 81 per cent of smartphone owners were aware of at least one application that incorporates machine learning. Usage, of at least a single application, was also strong, with about two-thirds of UK adults using one. 

The best known and most used machine learning-based application was predictive text. Two-thirds of respondents were aware of this application, and almost half used it which is understandable, given text entry is a core smartphone application. 

The second best known was the voice assistant. Forty-five per cent of respondents were aware of voice assistants but only 14 per cent used them. 
While the voice assistant was the second best known application, the second most widely used was route suggestions. Sixteen per cent used this, and 29 per cent were aware of it. Location-based app suggestions were less popular, with only 7 per cent using the application and a quarter aware of it. 

Over the coming years we would expect machine learning to become an increasingly important enabler and accelerant of smartphone usage, and a core differentiator of operating systems, specific phone models and smartphone applications.