With Machine Learning and Artificial Intelligence quickly becoming vehicles for industries to accelerate their growth, the number of students and young people taking an interest in these fields has shot up. AI is a concept to create intelligent machines that can simulate human thinking capability and behaviour. Machine learning is an application or subset of AI that allows machines to learn from data without being programmed explicitly. You can start your journey in learning how to wield these tools from the comfort of your own home because all you need is a working computer, an understanding of how these work, and an internet connection to look up problem statements and datasets to experiment on.
That said, these are complex subjects, and the problems are real problems, which can lead you down numerous rabbit holes before you find the right way to approach them. I jumped on the ML-AI bandwagon after realizing I found Natural language Processing interesting. It was a bridge between the English language and coding - both fields I was passionate about. I did online courses and then found myself interning in the AI-ML track at Pratham Education Foundation - an NGO focusing on educating children in rural and disadvantaged communities all across the country.
It was natural that the field of Education is tapped for exploring the possibilities of this technology, and Pratham has been, I daresay, one of the first to do so in ways that can benefit the entire system. The amount of data an open-school non-profit organisation like Pratham collects is astonishing, and their small but focused team of data scientists has taken on the onus of putting it to good use. There are multiple projects underway.
For example, I worked on automating the assessment of short free-text answers by children, which meant that after training on a considerable amount of children’s recorded answers along with indications on whether they were correct, the machine could evaluate new answers correctly. Another project was to create Knowledge graphs. These would condense chapters into a streamlined diagram of 5-10 interconnected keywords, so teachers knew the best way to navigate through the topic of, say, the Human Respiration System. These projects are, of course, all based on the premise that there will be resources available for the computational power these systems require when they come into use.
This nugget of information raised a flurry of questions in my head: Does India’s education system have the resources to implement these technologies in real-time?
The Covid-19 pandemic changed how children studied all over the world. It forced schools and educators to find new ways to keep children learning from their homes, on smartphones, tablets, laptops and even on the T.V, radio and through SMS messages. The pandemic presented a unique opportunity to examine how the old system of education needed to be adapted for modern problems, and to be in sync with new-age solutions. It highlighted the endless possibilities of e-learning. With so many children studying remotely because of the pandemic, did these advances in tech create a difference?
How can technology help?
1. Aiding Teachers
Innovative Artificial intelligence and Machine Learning systems can be used to facilitate and monitor children’s learning levels. AI can in no way replace teachers, especially where kids in their formative years of learning are involved. Soft skills and transferable skills such as critical thinking, problem-solving, teamwork and communication needs interaction with people.
However, AI-based teaching assistants and helpers can be provided for better enabling teachers to work with bigger groups. AI and ML systems can be used to automate the non-teaching parts of the job that come with being a teacher right now, such as grading papers, planning lessons, attendance records etc. A simple face-recognition system could be used to identify the pupils attending a class, or logging on for online lessons. Online/Remote proctoring for exams can be made easy with automated systems, and there won’t be a need for invigilators to be assigned to exam centres. The technology captures physical movements of examination candidates, and so, if a candidate tries to open a new window or URL it immediately sends an alert to the remote invigilator.
2. Simplifying Assessments
Adaptive assessments can be used to improve a child’s confidence in topics they aren’t good at and challenge them in areas they are. They can also be used to give personalized feedback for each child, and suggest how the child should go about learning topics they aren’t doing well in. The computerized test starts with asking a learner a mid-level question. Based on the response, the difficulty level of the questions in line gets modified. The application is powered by a large pool of questions drawn from data collected over the years, and the difficulty level is determined by the number of students who have answered the questions correctly. Adaptive assessments are designed in a way that they stop when a certain level of score precision or psychometric degree is met. Online systems of learning, which can only exist when there are devices available for each child to study on, will work at the pace of the child, not according to how much syllabus is to be covered for the year.
3. Reduction in Dropouts
Systems can be used to predict which students are most at risk of dropping out: based on previous scores, attendance and background systems can calculate and highlight students’ names to teachers, or pinpoint students whose performances can improve if given the correct support.
There are several EdTech organizations working towards making these systems a reality, and some are already widespread. However, for them to have a real impact, they need to be available and accessible to all children, everywhere in the country, which currently they are not. They promise a bright future, but shine a spotlight on the gap in the awning: internet connectivity.
We must acknowledge the mammoth in the room. Our education system is indeed a prehistoric being: large, and unevolved. In an ideal world, the transition from the classroom to remote learning would have taken place seamlessly, and all we would have struggled with would have been to figure out how to keep children engaged in the best ways even at home. The reality, however, is that we simply did not have the infrastructure ready for this change. The rapid shift to e-learning prompted by the pandemic has resurfaced long-standing issues of inequality and a digital divide in India that must be addressed by future economic, education and digitization policies. The World Economic Forum reports that 50% of India still does not have access to the internet - that’s 700 million people! According to UNICEF’S Remote Learning Reachability Report issued in August 2020, only 24% of Indian households have internet access to e-education. Internet connectivity for marginalized sections of society, for tribal and remote areas, and for lower-income households, has to be a priority for the government. The shutdown of schools hasn’t impacted people belonging to different tiers of economic means in the same manner.
Even before the pandemic, large scale learning assessments and surveys have consistently pointed to the poor learning levels of children even after eight years of elementary education. A critical factor impacting learning outcomes is the absence of a strong foundation provided by quality early childhood education. Another is the dearth of well-qualified and trained teachers. Even after recruitment, teacher absenteeism remains a concern due to poor governance. As outlined earlier, E-learning systems augmented by AI and ML serve as a solid solution to a number of these problems.
Is the expectation that waving the magic wand of technology will make all our problems disappear too great to harbour? It would be naive to do so, for sure. There is hope, however. The pandemic may have highlighted the digital learning divide in the country, but it has also accelerated research in the use of AI and ML in Education. We now need to develop systems that are either non-reliant on internet connectivity or find ways to provide resources that are adequate and accessible to all. Online learning will need to be recognized as a tool of the future, and shouldn’t be treated as the intermediate inconvenient solution to the shutdown of schools in the country and across the world. AI and ML systems cannot substitute teachers, but they can be a supplement, and the sooner we realize and embrace the changes these systems bring, the better prepared will we be for any future crises that occur.
About the Author
Khushi Goenka is a recently graduated Mechatronics Engineer from Manipal Institute of Technology where she was Editor-in-Chief of the Editorial board. She interned at Pratham's AI-ML track for a period of three months and was awarded a certificate of Excellence for her contribution. Khushi is the winner of Daimler Commercial Vehicles' Vigyaan 2019 innovation hackathon. Her interests are birding, baking, acting, and playing the harmonica.