Navigate Up
Sign In

Can AI help our learning?

Publication Date: Jun 13, 2019

AI article.JPG

Written by Mario Mallia Milanes

Mallia Milanes Mario.jpg
Without any doubt education is considered to be one of the pillars of a growing economy.  Governments and the electorate understand this fact all to clearly, but as times have change it is not easy to keep up with the demands and pressures of necessity. Mandatory schooling is not deemed sufficient to last a life time nowadays and life-long learning has become a necessity. Life-long learning is not meant to replace obligatory schooling, only as an augmentation to it.  When one looks at the training options available e-Learning comes close to none. 

It is attractive, cheaply available, flexible, is offered to a wide audience and is varied in content.  Educational institutions quickly got onto the band wagon as they saw e-learning as a potential revenue earner.  I see this from a different perspective.  E-Learning enables us to continue where we left off in school.  Sadly, trends show that despite many students enrol for such courses, relatively few honour their commitment to complete.  The reasons behind this phenomenon is varied.  In this short article we shall take a look at the issue and possibly offer one solution.

The Way it is

Many courses are available on a myriad of topics with equally varying prices too.  Most fall below the €100 mark, extending availability.  But this only attracts people to the course, it does not make them sit through it.  Looking at the content of such courses one quickly notices that the approach is a one-size-fits all strategy.  Produce a course and then sell it to many. The aim is governed by economics.  What about people who learn in different manners, and what about reinforcing your learning, and the collaborative aspect of people interacting in a class sharing experiences?  What about asking questions and investigating?  Living through your learning experience.


Learning can be broadly classified into four distinct ways.  Namely by:

a.     Association: where people build competences step by step;

b.    Active discovery: where one explores and learns as part of the process;

c.     Exchanging ideas with peers and getting support from them;

d.    Practicing skills with like-minded people.These theories proposed by academics to the likes of B.F. Skinner, Lev Vigotsky and Jean Piaget, have been around for decades and are entrenched in traditional teaching and learning methodologies.  In other words, they are tried and tested by educators all over the world. But somehow, they did not manage make the jump into the ether.

By closely observing the above approaches to learning we can make e-Learning courses better tuned to the student.  If one is to expect some form of return, other than financial gain, the user must naturally be consdered part of it. 

Bringing AI into the Mix

The theories outlined above suggest that education is far from a one-size-fits-all regime.  People have different aptitudes and tastes.  Out in the real-world adaptation of curricula has become more common place.  This because it has been felt that the different needs of students must be addressed.  Consequently, the digital world has to follow suit.  Primarily because e-Learning has to be of service to students and not a burden. So, we must find a way where teaching methods can be transposed to the digital realm without sacrificing flexibility. If current learning systems were to be augmented with the necessary interaction a learner might require, then the chances of him going through a whole course would be greatly improved.

AI article 2.JPGOne area that has been drawing a lot of research attention is that of Recommender Systems.  Recommender Systems are algorithms that are being used in a variety of scenarios to help suggest alternatives to people’s on-line buying choices or else help users make decisions. Amazon and YouTube or Netflix would be the place where we meet them the most.  We get suggestions that egg us on to make that extra purchase or give us ideas which would help us decide. 

Within an e-Learning scenario a Recommender System may be adapted to assist the student with different options that he may have to solve a problem.   Moreover, a new direction in research is currently trying to make the Recommender System explain to the user as to how it arrived at the results being proposed. Hence the term Explainable Recommender Systems.  This type of algorithm would assist a prospective student in following the line of recommendation.  In addition to this the user can also refine results by intervening in the decision process of the algorithm.

Taking A Closer Look

Recommender Systems have to base their suggestions on some sort of profile of the user or target data.  To do this the Recommender System has to draw up on vast amounts of data pertaining to user and topic profiles.  This would be considered a specific data mining application where the Recommender system would have Machine Learning capabilities enabling it to detect patterns within data and help users draw conclusions about those patterns.

AI article 3.JPGThe classical model-based approach to Explainable Recommender System would be that of Latent Factor Models.  In this approach the algorithm would attempt to learn about hidden or missing factors in data model.  Matrix manipulation techniques such as Singular Value Decomposition or Non-Negative Matrix Factorisation are prevalent.  Recently researchers are taking a keen interest in deep learning and representation learning techniques to solve ranking and paring problems.  The use of neural networks as a classification tool has once again become the preferred weapon of choice.

The idea I am trying to propose with this little article is that tools can definitely assist us in stretching the ever-elastic boundaries of knowledge.  To do this one has to keep the user, in our case the student, in mind.  Education must fit like a glove to its consumers, otherwise we will still get people who fall by the wayside.  An Explainable Recommender System my offer the possibility of making this goal possible.