AI and education
In popular imagination, Artificial Intelligence (AI) is seen as transforming the education system by focusing on developing learner-centric methods. Indeed, the past decade has witnessed schools shifting from conventional teaching methods to smart education with the aim of augmenting students’ learning experiences. This overhaul has been further fuelled by the present pandemic caused by the novel coronavirus.
COVID-19 necessitated schools to swiftly shift to online platforms for uninterrupted teaching and learning. While the disease itself created a havoc globally, it opened new avenues for creation and dissemination of knowledge through digital platforms. The proliferation of smart digital tools and data science in education system has provided many dividends. But what constitutes AI and how much of AI is really being used in educational technology? A closer look at what is currently deployed in smart education vis-a-vis the complexity of state-of-the-art AI technologies is needed to prepare a road map for future.
Education plays a critical role in moulding the future of society. In order to leverage the unprecedented advancements in information and communication technology, modern education systems need re-creation and revamping of content and delivery methods. Educational technology (Edutech) firms in India have initiated the use of smart tools to change the ways in which teachers teach and the ways in which students learn. Teaching-learning methods are undergoing tremendous changes as these firms are augmenting human learning with computer-based learning. This approach is trending because they are tapping on "personalised learning". With an increasing number of students, teachers may find it difficult to focus on each and every student on a continuous basis. Further, different students have different learning styles and they learn at a different pace. Thus, the firms believe that a one-size-fits-all learning approach is not the best method of education.
Personalised learning delivers content that specifically caters to the needs, preferences and competencies of individual learners. In collaboration with teachers and subject matter experts, these firms create a comprehensive database of questions, animated videos, quizzes, and related weblinks. It is in this personalisation of learning that machine learning algorithms play a critical role. Based on the level of the student, computer algorithms give questions to the student. From the correctness of the response, the system identifies the student’s strong and weak areas. Accordingly, it recommends topics, videos, questions and weblinks to address the student’s problem areas, thus personalising the learning path of the student.
How does the role of the teacher change? Smart tools augment the education system as the teacher is relieved from manual repetitive tasks such as taking attendance and tracking student’s daily performance. Teachers can track the progress of the students through a digital dashboard while devoting more time to develop critical and logical thinking among students. The teacher can focus on developing the skills of questioning and reasoning from an early stage of life. In addition, teachers can train students on soft skills and life skills including communication, empathy, social and leadership skills. Teachers can spend their time and energy in honing the personality of students and in developing a values-driven mindset.
Is personalised learning really driven by AI? While edutech firms claim that they use AI, experts in the domain of computer science and AI believe that the personalised learning approach can at best be considered as the starting point of data science. This is because edutech firms follow rule-based algorithms by training their application on historical data to develop personalised responses and unless the algorithms are retrained, the personalisation remains the same. Currently, these firms are doing well in the domains of rote learning and in handling mundane or repetitive queries through conversational tools and personalisation engines. But these systems work well in a very narrow set-up as concepts, questions, and answers in a specific chapter are limited. Thus, most edutech firms are in the infancy stages of data science.
Can true AI ever find its way in education system? The smart education platforms have been cruising on the success of true AI systems. However, technologically speaking, these different platforms sitting under the same AI umbrella tend to vary in their level of sophistication by gargantuan amounts. In popular imagination, these differences are hardly emphasised. Thus, IBM’s Deep Blue beating the world chess champion Garry Kasparov in 1996 is spoken in the same breath as the smart-tool educational platforms assisting teachers and students. But is there a fundamental underlying difference? One may as well ask if there is a fundamental difference between a Deepavali rocket and Apollo 17? Let’s have a closer look at AI technologies.
Humans have an ability to think, reason, learn and acquire knowledge from experience, whereas computers learn through algorithms. DeepMind’s AlphaGo (AI-program) defeated the Go-World Champion Lee Se-dol in 2016 who decided to retire from professional play after declaring AI invincible. While DeepBlue and AlphaGo were trained on the millions of possible moves as played by the grandmasters, a quantum leap was attained by DeepMind when in 2017 AlphaGo Zero ‘trained itself’ in a few days starting from scratch with random moves, most importantly, without any human input of historical data. While it is mind-boggling to note that AI has the ability to self-train and has reached superhuman level, it is critical to understand the power used by AlphaGo. AlphaGo used 1 megawatt of power as compared to a mere 20 watt of power used by Lee — 50,000 times more power than the human brain of the world champion. A journal article published in Nature highlighted that 1920 central processing units (CPUs), 280 graphics processing units (GPUs), and the brains of more than 100 scientists were behind AlphaGo that played against Lee’s brain (and his 1 mug of coffee!) The point here is that while we are awestruck by the abilities of AI, we must also be aware of the massive computation power demanded by these computer algorithms, thus making them not-so-easily available for the mass market.
What is the AI-way forward for the edutech firms? It is apparent that they are nowhere close to using cutting-edge AI technologies, even as they are not hesitant about neatly packaging smart digital tools as AI products. None of the personalised learning platforms are driven by self-learning algorithms. And looks apart, a robotic teacher is a well dressed-up virtual voice assistant whose abilities to engage in in-depth conversations with students are severely limited. In some ways, this is a part and parcel of our current times — as the Dalai Lama says, "there is much in the window but little in the room". Technologically speaking, Edutech firms are still at a superficial level of AI but these firms can continue to mine deeper into realms of possibilities. The past few years have seen exponential advances in the domains of image processing, pattern recognition, and natural language processing. Further, the use of technologies such as face recognition, eye-movement detection and emotion recognition can provide cues with regards to the attention levels of student. Physiological signals, speech recognition, and non-verbal behaviours captured through intelligent systems can provide insights into the level of understanding of students. Using this data, teachers can provide personalised course correction to students. These advanced AI-technologies that have immense potential to improve the teaching-learning process are yet untapped by the edutech firms.
While advanced machine learning techniques coupled with rich data has brought AI closer to human intelligence in certain ways, a true application of AI requires that the algorithms ‘self-learn’ from the data. The question remains, "Will Edutech firms be able to tap on to such sophisticated AI-technologies?"