Over the past year, many students and teachers across the globe swapped the classroom for remote learning and digital classroom meet-ups. Last April, WEFreported that over 1.2 billion children were out of the classroom globally; a trend that resulted in the significant rise of e-learning on digital platforms around the world.
Cloud-based technologies made remote learning possible. However, it was the use of specific technologies such as artificial intelligence and machine learning which made distanced learning more economical, practical, and easier to coordinate.
How AI has Helped Teachers and Students with Remote Learning
We spoke to Lead Data Scientist Andrew Poulton to learn more about the critical role that AI, and in particular, Natural Language Processing has played in helping teachers and students move to remote learning.
What does ‘EdTech’ mean to you, in 2021?
“In general, EdTech is the combination of education and technology, be that conferencing tools such as remote classrooms, or using technology to monitor progress. Over the past year, there has been an imminent need to offer distanced learning options.”
What have you been working on over the past year?
“An obvious use of tech within education would be for the grading of assignments and exams. There is a huge amount of work for teachers to do themselves in this area, so if you had a system that could reliably do that for you, this saves a huge amount of time.
I have been looking at ways to make this more reliable and transparent into why a given answer is right or wrong. You are never going to get 100% accuracy with these sorts of things; it may misclassify something as right or wrong sometimes. However, if the system can be transparent as to why it thought that, then it can help you to identify when an answer has been misclassified, and it will help you debug it and fix it for the future.”
Can you tell us more about the use of NLP in EdTech?
“There are lots of places you can use it. One obvious use case is in grading. When you have written responses, not just multiple choices, you can use NLP to work out whether the written answer is right or wrong. Then there are also uses in question generation. Here, from a portion of text you can generate a relevant and related question for students.
Another important aspect for education purposes is where there are models of how students learn, like a hierarchy of cognitive dimensions. These are important to building curricular assessments and different sorts of content for students. It can be useful for identifying what cognitive dimension you are testing e.g. ‘is this question testing the student’s recall, or their reasoning abilities, or their ability to apply their knowledge to an unseen scenario’?”
What are the current benefits and limitations of AI in education?
"Generally, it can automate a lot of things and make a lot of things available to people that would not otherwise be. It can also reduce the workload of teachers too as it can help them save time on things such as grading, which can be automated. This way, they can spend more time helping children who are struggling as they are spending less time on the tasks that can be automated with the help of NLP tools and AI.
As for limitations, you are never going to get a silver bullet for everything. At this point in time, the limitations are to do with the limitations of AI bias in general. Because of how important education is for students, there are all the sorts of external aspects that need to be taken into consideration.
An example of this is last year where the Department of Education applied a formula to automate and predict students’ grades and progress. It adversely affected students from disadvantaged backgrounds. This is an example of AI bias, and the biases that need to be accounted for, especially in areas such as education and healthcare."
Looking for a role, check our Xcede’s latest permanent and contract jobs in AI today.
