With increasing ability to learn & become better at assessing & assisting a student, AI based systems may not just be complementary to a teacher but can virtually take over some of the clearly definable outcomes demanded of a teacher, says Gokul Janga
With an advancement of technology through Artificial Intelligence (AI) that taps the ability of machines to mimic human thought and thought-based activity, there is a wide range of tools that can make the teaching-learning process a lot more fun and effective.
From Machine Learning (ML), computer vision or image processing to robotics, natural language processing, neural networks and deep learning, AI encompasses many specific areas of research and application that can play a big role in revamping the education sector.
Though the sci-fi movie level of expectations of the public may take a long time to be realised, AI is already a part of the daily life of many. There is general agreement that not a single area of human endeavour — making food, playing games, buying stocks or finding a mate — will be left untouched by AI in the very near future.
MACHINE LEARNING
Of the individual fields that make up AI, machine learning (MI) can be considered as the most representative of what AI means to the general public as it ties together developments in all other fields. ML is primarily about the ability of computers to achieve capabilities without being explicitly programmed for them.
These capabilities are achieved through algorithms that can themselves be categorised as Supervised Learning and Unsupervised Learning algorithms. Supervised Learning is where ‘labelled’ data whose properties are known, are used to train an algorithm so that the system can become better at classification and prediction and so can answer specific questions — like what kind of score can this student expect to score in a competitive exam or what value will a particular stock market index have in the next year.
Unsupervised Learning is where the system is trained to discover answers without even questions, by analysing the data and making its own inferences.
For example, when given data about previous exams, a system may be able to figure out that certain topics are the most important to crack that exam.
Most applications of ML that currently touch our lives (spam detection in emails, product recommendations, face recognition, medical diagnosis using images, creditworthiness, self driving cars) are combinations of the above types of algorithms.
Impact of AI
In Education, a critical area that affects current and future generations, we might yet see the most impactful application of AI.
An AI Tutor who has limitless knowledge, near infinite ways to explain a concept, adapts to every possible student personality, learning pace and is able to deliver any desired learning outcome may be a distant ideal, but edtech companies craft a Personal Learning Assistant for exams, are exploring paths to a practical result by investing in ML algorithms.
Personalised learning
How does a student read and for how long?, when does she read?, what does she read?, how does she answer a question?, what does the student get right and what does she get wrong when answering?, what topics does she like and what does she avoid?
By collecting data about a student, we can train an algorithm to classify the student, predict scores and customise a learning path for the student to achieve her desired goal. This goal, can be a rank in a competitive exam like JEE, a specific percentage of marks in a board exam, or a desired level of demonstrable competence in a topic.
Since the algorithm can be trained with ever increasing data, to become better, it is very achievable that every student anywhere, regardless of their pace of understanding, interest levels and goals can be provided a personalised learning path that maximises their potential.
Teaching through AI
Teachers, with the help of AI, can identify much faster the challenges being faced by students when learning and can devise ways to address them. Using AI, certain time consuming tasks like generating and grading of assessments, homeworks can be automated, allowing more time to the tutor for individual interactions.
AI can also strengthen the teacher’s ability to explain a complex concept by identifying the intermediary concepts that are needed to be understood and automating ways to achieve that understanding.
But, with increasing ability to ‘learn’ and thus become better at assessing and assisting a student, AI based systems may not just be complementary to a teacher but can virtually take over some of the clearly definable outcomes demanded of a teacher — like exam scores.
This is not necessarily a bad thing though, because a teacher will then be free, and in fact mandated to move beyond goal focused education that only emphasizes excellence in exams and other possibly inadequate demonstrations of competency.
It would be a singular success of the application of AI in Education if it were to make certain learning goals. Like specific levels of competency in subjects and their application towards scores in competitive exams, achievable with affordability and scale.
This will allow teachers to focus their energies on valuable responsibilities of engendering curiosity, inculcating excitement about learning, facilitating the development of critical thinking and well rounded problem solving skills, encouraging empathy, communication and a better understanding of the student’s strengths and weaknesses and thus overall helping realise a better human being.
The writer is CTO and co founderof HashLearn