AI Teaching Assistants, a Possibility or a Necessity?

Author(s): Ryan Lynch

Originally published on Towards AI.

Any student can tell you about a time (usually a few times) they’ve tried to catch a lecturer during office hours and couldn’t, or about an email they sent with a question last semester that still hasn’t been replied to.

And any lecturer can tell you about their jam-packed inboxes filled with questions about course material or the test coming up soon or students chancing their arm looking for some exam hints.

With the number of new entrants into third level institutions rising in recent years (at least in Ireland), this is becoming a bigger and bigger issue. For a long while now, lecturers have tried to handle this increase by hiring teaching assistants and tutors and lab demonstrators but this just isn’t scalable. It’s worth remembering these are usually all students too; students who have to keep on top of their own work and can’t fully dedicate themselves to helping students all year round.

Why is this such a big deal?

With the way most degrees are structured, students usually take 6 modules per semester, or 12 per year. Say if even just half of those have lecturers who don’t have the time to answer most basic questions in their classes, that leaves students with most of their time outside of classes spent chasing information like how to format their assignments instead of actually doing the assignments or studying.

So what actually is it that I’m suggesting?

Nothing life-changing, I promise. But damn useful all the same.

We’re all used to digital assistants in our lives at this point and chatbots are everywhere. There are the big names in voice assistants like Alexa and Google Home (Cortana never really took off for Windows) and there are chatbots for nearly everything; HealthTap answers users medical questions, Hipmunk lets people book flights, hotels etc through Messenger and then there’s InsomnoBot 3000 who keeps insomniacs company while they’re awake at night. So, is it a stretch to say we could use one as a teaching assistant?

Imagine instant replies to simple student queries, the time it can save for everyone involved. Lecturers can spend more time answering the tougher, more complicated questions without leaving others without an answer. Students can get more work done because they’re not waiting on someone else to help them sort their smaller issues that are getting in the way.

Feasibility

I can hear you saying “Sounds good Ryan, would it be hard to do?”

Thanks Reader, and no it wouldn’t be too hard. Really it depends on how complicated you want to make it. For a simple chatbot, it would be simple enough to create. With Oracle’s Digital Assistant you can build a basic digital assistant within an hour and a much more complicated one in not much longer than that.

And there you go again with the questions, “But how would you make it work?”

And now you’ll see the reason why I mentioned Oracle’s Digital Assistant; it’s built for easy and flexible integration. It already works with messaging apps so you can have your bot interact through Facebook Messenger (which most students already have) with absolutely zero hassle. And for getting assignment information? Most colleges use some form of virtual classroom to give assignments and make notes available to students. Say your college is like mine and uses Moodle, it’s completely open-source so you can use create an integrated system with it from information all online. What if your college uses Google Classroom? Still not a problem, Google Classroom has a publicly available API especially for this type of integration and development.

The other huge advantage of using Oracle’s Digital Assistant is that it already has Natural Language Processing (NLP) built-in. This means that you can set it up with your virtual classroom, some basic questions you’re expecting to be asked and then let Oracle handle the really complicated job of understanding the actual phrasing used by the students. NLP allows the system to understand the meaning of the words and sentences to get the actual meaning of what is asked, a task that has always been difficult for computers to do but which is very important for interacting with humans because we don’t always word things the exact same way as each other.

For example, NLP with a chatbot would ideally be able to tell that the messages “What assignment do I have coming up?”, “Tell me what’s due soon” and “Do I have any assignments?” should all return the same thing: a list of assignments and their due dates.

Personal Touch

One problem that I can see a lot of people assuming with this digital teaching assistant is that it might be taking away the personal attention a lecturer gives students in need. This is an understandable concern, with the lack of personal relationship with lecturers and tutors being one of the top reasons for a student to drop out of third-level education. But while it’s an understandable concern to have, it’s unsubstantiated.

By taking away a lot of the smaller, more mundane problems that lecturers need to help students with, you would be freeing up a lot of a lecturer’s time to interact with and help those at-risk students. Instead of trying to divide up their attention, they can make sure those who really need the assistance and the personal element are getting just that.

In Summary — It’s a Great Idea!

Essentially, by employing the use of a digital teaching assistant, all parties in the third level educational system are benefitting. Students with simple queries get a faster (or even just any) response; students with tougher issues get more focus from lecturers/tutors who have more time to give them; tutors can help students and balance their own work with the new time that’s been freed up; lecturers can dedicate themselves to helping students with tougher issues and to their research instead of just constantly replying to mundane, repetitive issues; the institution itself benefits by not having to hire more and more tutors and from their lecturers having better job satisfaction. An all-round win really.

If you want to give this a go, or you think you might have gotten inspiration for another great idea, have a look at Oracle’s free trial right here.

* All views are my own and not that of Oracle *

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI