Last time we talked about Artificial Intelligence (AI) in education, we’ve seen how smart speakers can help university students navigate campus life. However, voice-assisted technology isn’t the only one making a splash in universities; text-based communication is also a big hit.
It’s no secret that younger generations prefer text-based communication and they expect to find accurate information whenever and wherever they might be. It’s also natural that they have many questions, which they would probably not feel comfortable asking a university staff member such as “what do I do if I lock myself out of my dorm room?”. Additionally, they would probably need to ask this question at night when nobody is at the office anyway.
The ones leading this innovative process are the universities themselves, looking to solve problems such as the immense workload that comes with enrolling a large number of students into classes.
What 5 chatbots tell us about the future of HE
Today we’ll have a look at five great examples of chatbots in Higher Ed and how they fulfill an essential role at their respective universities:
Help with the admissions process: Lola
When The University of Murcia from Spain needed to make the admissions process a much smoother one, it enlisted Lola’s help. Despite the human name, which is an homage to a former employee of the university, Lola isn’t a person, but a chatbot.
Unlike a real person, Lola proved to handle the heavy workload of answering students’ questions, especially outside office hours. So much so that it was able to answer over 38,708 items with a more than 90% accuracy rate. Prospective students were able to ask anything regarding the admissions and the programs offered by the university.
What’s more, Andrés Pedreño Muñoz, the president of IT&IS, who was involved in the development of Lola, said that the chatbot “has enormous value because it can answer students’ questions using a natural language.” The emphasis on natural language isn’t surprising, as not feeling as if you’re talking to a machine is an integral part of the process.
Transitioning to college life: Pounce
Going from high school to college isn’t easy, as there are many steps to complete for enrollment in classes. Plus, motivation can seriously dwindle during the summer months, so much so that there’s a “summer melt” phenomenon, in which students don’t show up at all.
To counteract this, Georgia State University used Pounce, a chatbot, to assist students step-by-step. They’ve also kept track of the results. The AI answered more than 2,000 questions about various subjects, from financial aid to housing, which reduced summer melt by a whopping 21%. Students who used Pounce also had higher completion rates of key enrollment steps than those who didn’t.
The bot also uses personalized data, so it can proactively nudge students to complete tasks, based on their particular case. For example, students who have already completed a task, such as requesting financial aid, don’t get redundant messages to encourage them to do that. In this way, once they start asking questions, they’ll remember that they need help in various ways. The bot essentially opens a conversation that motivates them to do things on time.
Teaching assistant: Jill Watson
Jill Watson is notorious by now as a prime example of how AI teaching assistants could help actual teachers do their jobs better by easing their workload. So much so that at first, students were not aware that Jill, who could answer most of their questions related to the Knowledge-Based Artificial Intelligence class, was, in fact, not human. That was back in 2016 when professor Ashok K. Goel, Ph.D. and his team invested thousands of hours in its creation.
Even more, even if they created Jill specifically for the AI course, now it can answer questions related to other courses at the Georgia Institute of Technology. The team introduced Agent Smith, a bot that essentially clones Jill to “mold” itself according to a different area of study in about ten hours.
Most importantly, teachers spend a lot of time each week on administrative tasks, whether they’re teaching face to face, online, or blended courses. An AI-based TA answers routine questions and offers help when needed so educators can focus on what is essential: teaching, researching, and improving their methods.
Read more: 4 Ways AI will be a great teaching assistant
Knowledge retention: Quizbot
One of the most effective practices for retaining factual knowledge is active recall. Students can make flashcards and use them to study on their own. They can also learn together with a friend... and that’s what Stanford University’s Quizbot is.
The chatbot helps students retain information by asking questions, so they have actual conversations about the learning material. Much like a study buddy, it also gives hints when students are unsure of the correct answer.
In a university study, students who used Quizbot retained 25% more information than those who used a flashcard app, which has a similar function. However, the study only had 36 participants, so the effect might have been much larger if the study had involved more students.
IT help desk: Martha
Navigating those first days as a freshman isn’t just about making friends and figuring out where your classes are supposed to take place. It’s also about figuring out how to get the Wifi password, what to do when you don’t know how to log in to your learning management system, etc.
George Washington University has taken care of all IT-related questions with Martha, who first-year students interacted with 4,581 times during the pilot period. As these questions are considered routine, Martha frees up time for staff to deal with more pressing issues.
This is an excellent example of routine questions that don’t need human intervention as long as the system is good enough. Plus, students don’t have to wait until the next day when the IT office opens to find a solution to their tech-related problems.
Considering that Jill Watson was first introduced in 2016, the use of chatbots in Higher Ed is still a new trend. Despite the fantastic progress and good results, professors and researchers know that there’s always room for improvement. For example, institutions don’t have just one problem area, so in the future, there might be a need for a chatbot that can answer a myriad of questions encompassing admissions, summer melt programs, IT support, etc.
There’s also the problem of ensuring privacy. As an AI entity can gather a lot of data, it’s a matter of drawing the line between personalized help and respecting students’ right to privacy.
And there’s the conundrum of large scale adoption. As we’ve seen from our examples, the chatbots are a feature of a few universities, located in Western countries, who have the resources to invest in technology. Given enough time and progress, more institutions should be able to implement AI systems.
Even more, humans and AI, working side by side, will become a balancing act of knowing when AI help is most appropriate and when there is a real need for teacher-student interaction. For example, a human counselor is still most suitable for helping a student who has trouble choosing a career path.
That being said, integration with other systems such as a learning management system can be a real game-changer in personalizing learning. For example, it can remind students of assignment deadlines, help them check their grades, or set learning goals. There’s also a massive potential for adaptive learning assisted by AI, so the system can create a different curriculum for each student, recommend classes and learning materials, and simply help them make the most of their studies.
For help, ask Jill, Lola, Pounce...
The astounding progress of AI-powered chatbots promises to free up time for teachers, other university staff, and students. Chatbots are great for answering routine questions and show great potential in personalized learning so that all students can have a great experience. Plus, it’s a great way to ensure that future generations will be prepared for the AI revolution.