SDS 234: Data Science in Education

Welcome back to the SuperDataScience Podcast, ladies and gentlemen, super excited to have you back on the show here again today. And this time we’re continuing our series of FiveMinuteFriday episodes on data science in different industries.

So far we’ve covered off quite a few different industries including healthcare, mining, retail and energy. And today we’re talking about education. So as usual, we’ll look at five different applications of data science and artificial intelligence in this industry. And just as a reminder, the purpose of these episodes, of these series of episodes within the FiveMinuteFriday episodes is to give you some inspiration. So if you’re already in this sector, if you’re in the education sector, then you will find out some possible applications of data science in your sector. If you’re not yet in education, this can spike some creativity for you in your sector or you might be inspired to one day work on a project in the space of education.

So here we go, off to the applications of data science and AI in education.

All right, application number one, individualized learning. So when you were in school, high school or university or college, did you ever feel like you are ahead of the class in some subjects and maybe behind the class in certain other subjects?

That is totally normal and the reason for that is that the curriculum is designed with the bell curve distribution in the mind. There’s always going to be, in a large group of students which can be from 20 to 30 to a hundred or even 200 students, there’s always going to be a distribution in terms of how quickly people retain information and how quickly they grasp the concept. Some people are going to be around the mean with one standard deviation to the left to the right, that’s about 68 percent. But on both sides you’ll always have those 16 percent on the left, sixteen percent on the right so people who are advancing faster than the bulk of the distribution and people who are advancing slower than the bulk distribution. And when you’re designing curriculum for one class as a whole or for even a whole semester or multiple semesters, there’s just nothing you can really do about that.

However, now with the advancement of artificial intelligence and data science, we are able to tailor education better to individuals. And here’s an example. The average success rate for remedial or standard math courses is about 33 percent. Carnegie learning has developed a system using AI that aims to coach students on an individualized basis. And a randomized trial of 18,000 students showed a nearly doubled growth in performance on standardized tests, already in the second year of implementation. How crazy is that? So AI and data science can help almost double the performance on tests and basically the retention and understanding of material in a course such as, a complex course such as mathematics.

All right, application number two, automatic grading and teacher assistants. So automatic grading systems for simple multiple choice questions and answers have been around at least since the nineties. But with the advancement of AI, we can now grade even more complex questions and answers quickly, efficiently and without bias. Artificial intelligence, can’t really understand things like art but it can already, recognize spelling errors, incorrect structure, identify keywords and so on. And plus, on top of that, of course we’ve got plagiarism checkers. I remember when I was studying at university, I had to submit my assignments through a plagiarism checker every single time would check how my work is similar to other works that exist out there. And through machine learning, these plagiarism checkers have become more and more accurate at detecting cheating and other forms of plagiarism. An example here is that edX is a combined initiative between Harvard and MIT and uses an automatic grade assistant to grade essays. It started with teachers’ input, but has been learning since and has 11 major universities contributing towards its development. Research in 2016 showed a 95 percent agreement between the grades produced by humans and by the artificial intelligence.

So as you can imagine, that can massively help educators, lecturers, tutors to free up their time to actually focus on the creative components of education rather than just simply grading papers on and on and on again. I remember when I was at junior myself back into my bachelor’s, I would participate in exercises. It was compulsory for us to grade papers of newer students and that was a very, very tedious exercise. So this is some exciting developments happening in the space.

Okay. Application number three, improving courses. So sometimes when a teacher releases a course, and I know this from personal experience, it might be great at the start, it might be relevant, but with time it might, things might change in the world and parts of the course need updating. And when you have multiple courses or when you’re teaching multiple courses, whether online or at a university, it’s kind of hard to keep track of these things.

And moreover, sometimes when you release a course, there might be some sort of error that you made along the way. Especially if it’s a long course or you’re explaining something that’s obvious to you in a non-obvious way to people who are hearing for the first time. Well, artificial intelligence can help with that. And an example here is that Coursera is already putting this into practice, when it finds that or when a large number of students are found to submit the wrong answer to a homework assignment, the system alerts the teacher and gives future students a customized message that offers hints to the correct answer. So basically data science can help track how the course itself is actually doing, based on the performance of students overall in the course. So if there’s a large deviation in the performance of students, maybe it’s not the students’ fault, maybe something’s wrong with the course and that is used as a flat.

Use case number four, improving graduation rates.

So despite of what you might have heard from different various sources, the actual goal of colleges and universities is to help people graduate. They’re not there to fail students on purpose. They want to help. They want to make sure as many students as possible graduate. However, that’s of course not the case. There’s lots of students that don’t end up graduating, so universities want to look for ways to increase these graduation percentages and decrease the number of people who drop out of college, drop out of university. And while that is feasible in a kind of manual way when you have a small number of students and you can do that through trial and error, when you come to a large college level with thousands of students, where for example, Stanford has 17,000 students, Oxford 22,000 students, Harvard 23,000 students, the University of Queensland where I studied has 50,000 students. When you’re talking about a large cohorts of students like that, it is virtually impossible to do it manually.

And that’s where artificial intelligence can come in handy. And an example of that is the Temple University in Philadelphia which has about 40,000 students used data science to figure out which students are most likely to drop out and then offer them additional support. This increased their four year graduation rates from 20 percent to 44 percent and six year graduation rates from 59 to 70 percent. As you can see, that’s a massive improvement and that actually helps people finish what they started, finish their degrees.

And, finally application number five, artificial intelligence teachers.

So this is the next level of education where we will be learning not from human beings, but from artificial intelligence. So this use case is in its very juvenile state. It’s still very early days, but progress is already being made in this direction. And my favorite example is a Global XPRIZE which is run by Peter Diamandis, which is currently actually taking place. And you can actually find this XPRIZE. It’s called, you can go to xprize.org/prizes/global-learning. And this prize has been going on since September 1, 2014 and it will end on the 28th June, 2019. And so what is this all about? Well, $15 million dollars will be awarded to the team that is able to create an artificial intelligence learning system that follows certain criteria that you can find on the website. But basically in short, what the goal is is to help people around the world get education. So it’s estimated that about 250 million children, 250 million children around the world cannot read, write, or demonstrate basic arithmetic skills. And so what this prize is designed to do is, or what the goal is, is to create a system where, for instance, you can put this application with artificial intelligence or however it works, onto an iPad and then just leave these iPads in various villages throughout Africa.

And so the criteria, one of the criteria is that a child is supposed to be able to pick up this iPad and to be able to start learning, without any kind of supervision. Just by clicking on the iPad, the AI or the system in place should be able to guide the child through the learning process and teach them things like, English or French or other languages, mathematics, geometry, writing skills and things like that. So we already have the technology in the sense of hardware, like indeed we can just leave an iPad and it will work. It doesn’t really matter where in what part of the world it is. You can always, as long as there’s electricity, you can always go to charge it up. But the question is now how do we develop the artificial intelligence that is going to power this.

And so this XPRIZE, these 15 million dollars are designated towards that. Very interesting to watch to see how this is gonna all play out. So they’ve already done the finalists. Now they’re going to announce the winners in May, June this year. So very excited about that. But that’s one of the initial projects in that direction and after this XPRIZE, hopefully we’ll see more and more developments in the space of artificial intelligence teachers.

So there we go. That was five use cases of artificial intelligence and data science in the space of education. Of course there are plenty, plenty more use cases. Education is a very big industry, is about five percent of the world’s global GDP or in some countries like Scandinavia is actually eight percent, so it’s a very important niche and we also know it’s important because we’ve all gone through education ourselves in the fact that you’re here listening to this podcast still listening to this FiveMinuteFriday episode, towards the end means that you are all about your personal growth, personal education, and I’m sure some of these examples have been inspiring to you.

So on that note, thank you so much for being here today. I look forward to seeing you back here next time. And until then, happy analyzing.