Written by Dr. Sandeep Bansal on Digilah (Tech Thought Leadership)
Everyone knows that different individuals possess different capabilities, comprehension abilities, problem-solving skills, and hence the learning needs vary across the students of a class with their varying interests, abilities, performance, pre-requisite knowledge, etc.
In order to address such needs, the adaptive/Personalized learning/teaching systems are being worked upon with the involvement of technologies like Artificial Intelligence, Machine Learning etc. These systems provide different learning paths with different paces to different learners based on their needs & performance but they use the same study material for all the learners; though it may be delivered to them at a different pace depending on their needs and performance; while the learners need study material as per their need, capabilities for the same topic. This gives a rise to the need of not only developing different study materials varying with variations in different parameters but developing the machine learning models also which would take care of not only the learning needs being dealt with by today’s system but other needs being discussed further here.
Countries like India have lots of diversities with respect to language, culture, regions, etc. Such learning systems can play a key role to bring inclusivity in education as we note:
- Learners with physical disabilities (Divyang Jan): In India, around 1% of school-going children are children with physical disabilities and need transformation of e-content.
- Learners with learning disabilities: In India around 5% of the school going children are affected with learning disabilities (dyslexia, dyspraxia, dyscalculia and dysgraphia etc.. Each type of disorder may coexist with another). In such cases also the study material needs transformation.
- Learners with different socio-cultural identities, socio economic & geographical identities with the needs of study material in different languages, dialect & culture.
The fundamental principles of NATIONAL EDUCATION POLICY (2020) of Indian Government include –
- recognizing, identifying, and fostering the unique capabilities of each student
- focus on regular formative assessment for learning
- extensive use of technology in teaching and learning, removing language barriers,
- increasing access for Divyang students, and educational planning and management
Therefore the role of AI based adaptive system delivering the right content at the right time to the learners in personalized manner has a very important role to play. The machine learning models need training data also for their adaption to the different learner’s need with regard to their learning attributes (like some emerging system) & varying needs of different content for the same topic (not common & to be evolved). How it all would work may be understood with a simple architecture of the whole system. The basic building blocks including the content would be as follows:
- Artificial Intelligence-based Decision System: This would be the core of the whole system to interface with the front-end i.e. learner’s interface, with the learner’s repository where the learner’s attributes would be stored and with the content repository where the content of different languages, formats, and levels to choose from for the learners would be stored.
- Learner Interface: It provides the test material to the learners and based on performance captures different attributes of the learners e.g. learning disabilities, problem-solving skills, comprehension abilities, misconceptions, gaps in prerequisite knowledge which are then sent to the “Learner’s Repository” module. The whole process is controlled by the core AI-based Decision system. The interface may be equipped with conversational AI (powered by Natural Language Processing, Speech recognition to interpret the intent of the user and providing smooth interaction with the system) in his/her language.
- Content Repository: Different type of content e.g. Content for physical disabilities, Content for learning disabilities (the content of different levels of explanation with a provision of need-based detailing of basics & prerequisites involved in the concept to enable the learner to drill down), The content to address the needs of study material in different languages, dialect & culture etc.
- Learner’s module: The attributes of the learners captured by the AI system through the learner’s interface would be stored in this repository consisting of attributes of different learners with their respective learning attributes and learning paths to be followed by them.
Learning Path (sub module of learner’s module): To decide on the appropriate learning path for the learner, the system first evaluates the learners with respect to different attributes, different learning issues etc.. For example a learner first needs to understand the concepts of Motion to understand “Simple Harmonic Motion”, then waves, then Light’s concepts, then concepts of reflection & refraction and so on. During assessment of the learner, if it is found by the AI system that the learner is lacking somewhere, the system would advise the learner to go back to the basics.
Content Selection by the Decision system: The content selection logic of the system would take care of the content selection at all stages from the content repository based on different attributes & need of the learners. Based on various factors of the learner’s as summarised below, the AI based system would choose the appropriate content for the learner.
For the proposed system the availability of content for training the models & e-content for the learners in different formats is a bigger challenge. The initiatives of the Government of India e.g. Natural Language Technology Missions (targeting content in Indian languages), guidelines, policies for e-content development (including those for children with special needs) by Ministry of Education, Ministry of Social Justice & Empowerment would play a key role to make such content available. As an outcome a lot of content is expected to be made available in different languages & formats which may be used for such solutions.