Artificial Intelligence (AI) might be safely considered as one of the most promising megatrends of ICT research of the XXI century, given how wide is the array of its possible applications, such as business, healthcare and cybersecurity. The multitude of scenarios and practical situations where AI could be deployed make it a General Purpose Technology (GPT).
AI is a branch of computer science based on the production and deployment of state-of-the-art algorithms, each having the ultimate goal of mimicking in-depth learning processes, thereby enabling the production of new, valuable information, or enabling the user to perform some decisions. If well implemented, artificial intelligences can indeed provide many valuable outcomes, such as:
- making forecasts
- supporting decision-making processes made by humans
- generating objects (i.e. text, music, images etc.)
- recognising speech, text, images and so on
DEPLOYING AI IN THE E-LEARNING INDUSTRY
Since AI technologies are generalised in their applications, they are suitable to be deployed in any kind of business context, and therefore, for obvious reasons, they may fruitfully be applied to improve LMS (learning management system) for e-learning. e-Learning represents a recent trend in the learning industry, it mainly consists on providing remote learning solutions and courses via web platforms, pretty much spanning any field of interest from literature to science.
e-Learning has been a growing market in the last decade and it significantly evolved both in the quality of the services provided by specialised LMS platforms (with great results in terms of customer satisfaction over time) and in the quality of the courses.
WHY AI CAN IMPROVE E-LEARNING
Customising individual learning patterns and providing effective customer support are the cornerstones through which AI might play a role in its application to e-learning. LMS can indeed be organised in such a way that, by leveraging on AI solutions, it would be possible to increase the accuracy of customisation techniques, given a sufficiently broad and specific line of courses.
An additional feature of e-learning platforms consists in adaptive learning, i.e., a form of training orientation mechanism which helps the students to find the most interesting course in the platform given their knowledge basis or their academic background. With adaptive learning, also contents are managed and balanced, so that each student can really fit on his or her learning curve.
e-Learning platforms often provide courses and material for a multitude of fields of interest, and because of this, provided that the number of contents might be considerably high, some students would end up being confused and not able to orientate themselves correctly. This happens especially for people who are not experienced with the platform.
Orientation is not just a fancy add-on to the e-learning platform, it is a need for students; developing an intelligent orientation system might be a great source of value for the business running the platform. That’s a first and strikingly important opportunity to deploy Artificial Intelligence in e-learning. The underlying concept relies on embedding a certain kind of algorithm to be fed with data produced by the interaction of the student with the website, thus detecting topics of interest for him and outputting advices on next courses to follow.
Technically speaking, these results might be obtained by means of a recommendation system, which is a state-of-the-art unsupervised deep learning model, having the ability to create arrays of recommendations for the interacting users and help them making decisions (e.g., buying some product or, in our case, selecting a course to attend).
Using recommendation systems in online platforms is actually nothing new for many successful companies (such as Youtube and Amazon) and its application in e-learning field proves to be very useful. In order to be effective in the context of an LMS, a recommendation system would do better if fed with more data (in this case, course contents); this implies that a good platform should be endowed both with AI solutions and with a considerable amount of contents to provide.
Another aspect of fundamental importance is to offer assistance to the learner in difficulty. In this area, the application of artificial intelligence still finds a very interesting and useful research field.
The learner may need to have information on the individual courses or more generally on the use of the LMS platform.
E-LEARNING AND CHATBOTS
Chatbots are artificial intelligences based on semantics whose purpose is to interact with people visiting the website. The way a chatbot works stems from a certain combination of reinforcement learning algorithms (i.e., the training logic of proper, physical robots) and Natural Language Processing (NLP). Chatbots can elaborate both written text and speech, thus producing answers accordingly and, by means of that, holding a meaningful conversation with the user. To this last extent, imagining chatbots as an industry-standard for customer support is quite straightforward.
Chatbots are already used in LMS for technical customer support, however, the future challenge will be to turn them into “intelligent tutors” for attendants, i.e. programs which would be able to replicate the actions of a human teacher and to provide feedbacks and advice to learners. Although this shall be a complex operation, leveraging on big data will allow chatbots to interact in an “intelligent” way with users.
It should come as no surprise that chatbots are actually pretty much used by several businesses operating in many industries, however, the next challenge for the improvement of bots will be to train them in such a way that they could eventually enable customised support on single course contents, sourcing course-specific information and interacting with the learning student while he or she is attending, although this kind of technology would require quite big efforts to be implemented.
Conclusions
During the last two decades, Artificial intelligence has been growing and improving tremendously: what once was considered to be hard to implement (see Neural Networks and deep learning) is now produced on a mass scale and is continuously improved. There is no doubt that the future of AI and its applications is quite bright right now, but the main engine of growth will always be through businesses, especially the most scalable ones.