How machine learning is going to revolutionise education
Machine learning is going mainstream, and it is going to impact education. Does that mean we are all going to be taught by AI holograms now? Not necessarily. It seems that machine learning, in the first few years at least, will act as more of a support for human teachers than as a rival.
Firstly, though, what is machine learning? Simply put, it’s a computer algorithm that can program itself, based on feedback it gains from the results of its explorations. It makes data-driven decisions rather than following static program instructions. At first, a machine learning AI is like a small child. It gradually learns by trial and error in the same way a toddler might learn not to touch hot things.
But it learns quickly. Already it can predict heart attacks better than human doctors, beat world champions at chess and go, drive our cars for us and learn to make sense of our speech. It is scarily good at learning, and outpaces humans at many data analysis tasks. So how useful is it in education?
With Natural Language Processing, machine learning algorithms can grade not only multiple choice test answers but also free form written answers. It’s not quite ready for prime time, but in some cases it can outperform human graders.
eLearning company Blackboard recently announced a collaboration with the IBM Watson team on advanced analytics to analyse student online behaviour to help deliver personalised learning. The idea is to use machine learning AI to optimise adaptive learning pathways.
As well as personalising learning pathways, it can organise and modify the content itself by using content analytics to optimise content modules. Examples include Gooru and IBM Watson Content Analytics.
EdTech companies like New Classrooms are using machine learning for dynamic scheduling, matching students who need help with teachers who have time, based on learning analytics data.
Most applications of machine learning, then, are in taking time-consuming work off teachers’ hands, freeing them up to do the touchy-feely human stuff that AI can’t do yet. It can do humdrum back-office tasks like bus route scheduling and dietary analysis, too, to keep young brains properly nourished so they can learn.
Oh yes, and machine learning is also touted as a solution to cancer, climate change and terrorism.
Hype? Well, research firm Gartner’s recent hype cycle report does put machine learning at the peak of its curve, meaning it is at maximum ‘inflated expectations’. But it is already here and doing useful work in education. Machine learning should be top of everybody’s list as a technology to watch.