Blended learning will become an increasingly important methodology in delivering education at school level. There is already good research emerging that indicates that a blended approach can lead to greater student gains than either a purely traditional face-to-face or a purely delivered online course.

The definition of blended learning has evolved over the few years that it has been around, but is generally accepted now to mean a course of learning that includes some face-to-face (co-location) work and some distance work through an online element. Its success seems to make obvious sense; by adding in the distance learning element, students can continue (and are encouraged or required to) their learning beyond the usual timeframe. They can have time to reflect, and community elements of the online environment means that they can also act in the role of knowledge creator, contributing their own thoughts and ideas.

Blended learning is settling in to a definition that requires co-location and distance work. But what if, just as we are agreeing on a definition, the definition itself is about to become redundant?

There are different cognitive activities taking place when working face-to-face compared to distance individual work. It is about the human interaction, the ability for the learning to adapt to reactions, the being a part of a learning network, the access to external expertise (usually in the form of a teacher), the skill of the teacher to plot an infinite number of routes through the learning based on the reactions and emotions of the group, and the whole spectrum of human emotions that come with feeling a sense of togetherness with peers.

Suppose, then, all of these aspects could be replicated without co-location. Would that element of blended learning be necessary?

Increasingly we, and particularly children, are forming social and emotional relationships without face-to-face interaction. Social media and Web2.0 have enabled this. But add to the mix intelligent algorithms that can lead to true Learning Analytics and Adaptive Learning, then we could have an online element capable of high level artificial intelligence, able to play the role of the external expert.

Learning Analytics is not the same as data analysis. It is not tracking. It is not taking data from tests and the like and then spitting these out at a teacher to interpret. Learning Analytics is real-time, live analysis of the interactions that students are having with a learning environment. Imagine the thought processes that a teacher has when watching a child working on a problem – they tweak the experience, they prompt, guide and assist, based on what the child is doing and their own experience of observing hundreds of children attempting the same problem in the past. They have built up an understanding of the behavioural patterns and are able to predict where the child is heading and therefore how to best intervene.

Learning Analytics brings the same observations. And Adaptive Learning plays the role of guiding. Combined, if the algorithms measuring behaviours can be got right, Learning Analytics and Adaptive Learning can behave as an artificial intelligence.

The learning environment can operate just like a teacher with a class of varied students.

Co-location is now no longer necessary.

We're not there yet. The algorithms are really tricky. But as I explore early artificial intelligence – funded and driven largely by the gaming industry at present – I can see how close a truly intelligent platform might be. 2015, I'd say.

So perhaps in the very near future we might have to revisit the definition of blended learning once more and ask ourselves seriously whether or not co-location is a necessary ingredient anymore.