Machine learning and one to one tuition
Technology does anything but stand still. So just when we’re starting to get to grips with VLEs, now we have to talk about machine learning! If it sounds like I’m complaining, I’m not. Technological improvements, not just for their own sake but for the sake of easier learning a more well-educated population, are welcome in my book and anybody’s book. That’s why I thought that today, we could look at machine learning and how it could revolutionise education. Read on to find out more!
What is machine learning?
First things first, machine learning is just what it sounds like: software designed to be responsive and flexible to the needs of those using it. Essentially, machine learning involves software (or, in some cases, physical robots) that can respond to their environment without being told what to do by a human being.
An example of a robot that can ‘learn’ is one that can find its way out of a maze, being able to remember the turns it has already taken. An example of software is a program that can alter its responses regarding received data, like answers to set questions.
Machine learning takes everything that technology has been and tries to flip it on its head. Computers have always been inflexible: if the computer says no, the computer says no. But machine learning means that a computer that said ‘no’ could reflect on the data it is given. And upon reflection, the computer could change its answer to ‘yes.’ That’s why machine learning is at the very forefront of the AI revolution: because to change one’s mind is a fundamentally human, or at least animal, trait.
So far, efforts to create machines or programs that can learn are at a fundamentally early stage. If you’re familiar with anything along those lines, it’s probably the ‘chatbots’ you can easily find online, which are ‘taught’ by the internet users who interact with them. Like most similar things on the internet, it hasn’t ended well! But software engineers and technological visionaries are looking further and devising genuine uses for machine learning that could drastically improve the provision of education worldwide. 
What difference can machine learning make in education?
Even though we’re only in the early days of machine learning, it’s already starting to make a difference in the classroom. The only thing holding us back is a fear of innovation.
A great example is a rudimentary AI used in Indian schools called Mindspark, developed by a company called Educational Initiatives. Mindspark is a project more than a decade old, and its creators designed it to be flexible in response to the vast amount of data generated over that time. 
So, it can anticipate common errors in pupils’ homework: for example, it understands that a pupil’s belief that 3.27 is larger than 3.3 because of what’s called ‘whole number thinking’, where since the whole number 27 is larger than the whole number 3, 3.27 must be the larger number too. When it identifies this error, it can recommend a pattern of exercises that will help that pupil understand what’s at the root of their wrong answer. 
This is impressive enough, but Mindspark’s productivity increase is underlined by genuine scientific study. J-Pal, a group at MIT that studies poverty alleviation, checked out Mindspark at work. They found that in a group of Indian children who used Mindspark for just shy of five months in an after school class, their maths skills dramatically improved, as did their understanding of grammar and spelling. And all at a fraction of the cost of direct, human to human tuition. 
What’s coming next?
Mindspark is an excellent example of both the fantastic benefits and current flaws of AI and machine learning. While it excels at teaching pupils about maths and grammar, it can’t yet teach them to understand poetry or compose a song. The reason why is because we haven’t yet developed AI that can understand art or culture. These are fundamentally human concepts, which don’t necessarily appeal to reason or logic but human emotion. It’s difficult to imagine AI getting to grips with that any time soon.
Even so, the current level of AI and the rapid recent development of machine learning means that we are light years ahead of where we were ten years ago. Ten years ago, virtual learning environments started appearing, and twenty years ago, we started playing with white boards. There’s no telling where we could be in twenty years from now. Even if AI is always limited in its application, we can still improve how we teach maths and anything else based on laws easily understood by machines. That’s no bad thing.
A good indication, though, of where we’re heading is the dramatic personalisation not just of learning but of life. Children are increasingly empowered to choose the direction of their learning through VLEs and increasingly through technology like the tablets and laptops that are an increasingly common sight in schools.
Machine learning in schools: the final verdict
J-PAL’s study gives a good indication of what machine learning can do. And this is only the beginning. As artificial intelligence becomes more sophisticated, we will see more effective and efficient education, with an even more significant effect on students’ learning. Like VLEs, AI could potentially reduce the ever-expanding schedule that teachers are facing. So, on that front, machine learning is a win-win. The only problem is waiting for it to get even more effective and become more widely accepted.
Personalisation and digitisation are not without critics, though, who view them as a hindrance rather than a help. Allowing students to define how they learn, in turn, will enable them to determine the methods they use to find facts, and having Google or Bing at one’s fingertips at all times disincentives memorisation of facts and theories. It’s a psychological shortcut, one which we’re all guilty of but which will probably disproportionately affect those who grow up with it.
Sources: http://uk.businessinsider.com/researchers-predictions-future-artificial-intelligence-2015-10?r=US&IR=T  https://en.wikipedia.org/wiki/Mindspark  The Economist, July 22nd, 2017, pp18-20.