AI in Education - Tools for the Future of Higher-Ed

AI in Education: Tools for the Future of Higher-Ed

Universities face innumerable challenges when it comes to enrolling students, planning curriculum, teaching students in a personalized way, and countering high drop-out rates. When tackled manually, these challenges can be overwhelming. But bring in data analytics and AI in education and use them correctly, and you’ve found a way to promote institutional success. Armed with this knowledge, colleges and universities have created personalized chatbots, mined data of their applicants, and designed data-fed virtual teaching assistants. Several technological and educational leaders are now using AI in education to offload time-consuming, repetitive tasks, and deliver a better learning experience for students. It’s only time before the rest of them follow in their footsteps.

AI Applications

AI applications for higher-ed are designed to help universities and faculties enhance their strengths, and be helpful in guiding the decisions they make. Consequently, they can be divided into three types - Institutional, Student Support, and Instructional. While Institutional applications deal with how universities will interact with the incoming applicants, Student Support and Instructional applications help universities guide their students better. Let’s look at each one of them, one by one.

Institutional

Applicant Data Mining

A Wall Street Journal article, titled “Colleges Mine Data on their Applicants”, was published in 2019. It talked about how some universities were using machine learning and artificial intelligence to see which prospective students were genuinely interested in their programs. They tracked how these students interacted with the university website, engaged with their social media platforms, and how quickly they responded to emails. Over 80 such variables were considered to understand the response of each student. Then, the colleges and universities used the results from their analysis to determine the profiles of students they should reach out to, which topics to emphasize on, and which students to consider for admissions. They even chose whom to provide financial aid to, on the basis of their results.

Recommendation Engine

College applications, however, are a two-way street. Just as institutions are using AI to shortlist prospective students, students should be able to sort through universities and programs to find the ones that best suit their profile. iSchoolConnect (iSC), a higher-ed tech company, noticed this gap and filled it by building their Recommendation Engine.

The Engine accepts a set of parameters - like academics, test scores, co-curricular activities, and work experience - and generates a list of schools customized according to the student profile, making the shortlisting of universities a much easier task.

Curricula Planning

Speaking of customization, Knewton, an adaptive learning company, developed a platform to personalize educational content. Also known as Alta, their platform adapts to students’ proficiency levels with each interaction. It identifies the gaps in the students’ knowledge, helping them understand which areas they need to work on. Then, Alta proceeds to provide relevant coursework materials and makes sure students stay on track with their courses. The company has partnered with Arizona State University, University of Georgia, and has already seen commendable results from Johns Hopkins University. According to them, the platform not only helps students score better, but also improves the quality of their assignments and their ability to retain information.

Student Support

Assessment Center

Within a single month, iSC’s Assessment Center proctored close to 50,000 hours of tests. The solution was developed during the COVID-19 pandemic to help institutions conduct exams remotely. While students took tests from the comfort of their homes, the Assessment Center monitored their activities live. The AI kept a lookout for suspicious behavior - like gazing off-screen, talking on a phone, suspicious hand gestures, etcetera - and flagged it. Now, students had greater flexibility in booking test slots, and there was no need to hire more staff for invigilation. And this is only one of the things AI can do to reduce the workload of a university’s faculty and staff.

University Chatbot

Last year, the University of Murcia rolled out an AI-enabled Chatbot to attend to students’ queries about its programs, application procedures, and campus experience. The bot - Lola - spoke to more than four thousand students and answered questions correctly more than 91% of the time. Available even outside of office hours, Lola replied to repetitive questions in no time, allowing the university staff to attend to more productive endeavors. Soon, Staffordshire and Georgia Tech came up with their own chatbots, designed to answer all the frequently asked questions by students.

Now, several universities are reaching out to organizations that specialize in designing such chatbots. Take CASIE for example - a creation of iSchoolConnect, she answers all queries related to universities, their programs, tuition fee, how to go about applying, the documentation process, etc. And she does it with calculated wit.

It’s imperative to highlight that, along with saving time, these chatbots collected pools of data about what students were most interested in finding out. This not only helped in improving students’ experiences but also in offering new programs and planning the upcoming curriculum.

Smart text messaging

It’s one thing to recruit students, and another to make sure they don’t drop out. This is where Georgia State’s Smart Text Messenger - Pounce - came in. Before Pounce, a lot of students who had joined in spring would not show up when the fall term began. Knowing that it would be a phenomenal problem for university staff to tackle, the university partnered with AdmitHub and came up with a conversational AI text-messaging service for students. Soon, they realized that students weren’t showing up because they were confused about which forms they needed to fill, and didn’t know which office to go to have their queries answered. Armed with this knowledge, the university enabled Pounce to answer these questions 24/7 via text messaging. The service answered more than two hundred thousand queries over a single summer, tailoring its conversations to meet the needs of each student. Cutting out the hiring of 10 full-time text-messengers, Pounce reduced the student drop-out rate by 22 percent, resulting in an additional 324 students showing up next fall. The university continues to use its messaging service to date.

Instructional

Virtual Teaching Assistant

During a single semester, about 300 students of a class in Georgia Tech sent more than 10,000 messages to an online message board. A teaching assistant, Jill, was assigned to answer all their queries. And she did. Handling a volume of questions that is impossible for a regular teaching assistant to answer, Jill became their favorite. Her answers were fast, effective, and helpful. Why wouldn’t they be? Jill’s memory had been fed with tens of thousands of questions from the past semesters. Jill was a virtual teaching assistant. Over time, Jill learned to parse the context from the queries and answered them with increasing accuracy, reaching a success rate of 97%. Even though she was able to answer most of the questions correctly, what Jill couldn’t do was explain nuances and exceptions, help plan coursework, and motivate her students. This gave time for other teaching assistants to do more meaningful work.

Smart text messaging

Cognitive immersive classroom

More recently, IBM Research partnered with Rensselaer Polytechnic Institute to help students learn Mandarin. They paired an AI-powered assistant with an immersive classroom, naming it the Cognitive Immersion Room (CIR). When inside this room, students find themselves standing in a restaurant in China, a garden, or a Tai Chi class, where they practice speaking Mandarin with the AI assistant.

The CIR is only one of the 4 projects undertaken by the Cognitive and Immersive Systems Lab, where researchers are also working to augment group intelligence in real-world environments. This could fundamentally alter the way students learn.

Writing Mentor

AI tools not only help universities improve student experiences, but also help students get the best out of their education. Two years back, iSchoolConnect rolled out its Document Grader. Designed to assist students in writing better SOPs for college applications, the tool checks for sentence structure, grammar, and plagiarism. Based on it’s assessment, it grades an essay for its readability and subjective relevance.

Imagine, then, if one were to expand the subjectivity of this Document Grader outside of SOPs. The tool would not only help students get an admission, but also aid them in writing assignments and collating research as they continue to study. Extended to review videos instead of texts, this tool has even been transformed into a Video Interview Analyzer (VIA) by iSC. The VIA helps students enhance their ability to answer questions more deftly. It monitors and reviews their expressions, tone, and gestures as they answer interview questions, pointing out where they can perform better.

Challenges

From shortlisting and enrollment, to assistance and placement, AI has penetrated innumerable aspects of the higher ed industry. We’ve come far from when students had to wait for days to get their queries answered, or when university staff spent hours on mundane, repetitive tasks. Over the past decade, AI has transformed the way institutes recruit, teach, and engage their students. But the field does not come without its challenges.

To start with, students know that information about their lives and preferences is being used to make data-driven decisions. They are concerned about their privacy. To get through this, we need to have policies in place that ensure their data is being used transparently and ethically.

Students must be allowed to own and control the flow of information universities have about them.

Another concern is the way the data is being processed. It’s necessary to make sure the data used to make decisions for one group does not translate to a standard model that gets used for another, seemingly similar group. Keeping bias out of our machine learning models is difficult, but imperative.

The third and final challenge we face is the lack of technical skills in the university staff. The projects we talked about were successful only because they were built using huge amounts of data. This implies that, in the beginning, someone has to feed data to a system so it can eventually start scaling and function well. Therefore, the individuals who do the data feeding have to be tech savvy when it comes to knowing how to use AI tools.

Conclusion:-

As educational institutions start exploring the benefits of using AI, mindless and routine tasks will be automated. Against the popular fear that AI will replace human beings and deprive them of their jobs, artificial intelligence is meant to enhance human capabilities. AI will not only help the higher ed industry transform student experiences, but also free up the university staff to work on and solve more demanding problems. It is essential to remember this, then - the best results will come when we combine the strengths of AI with human ability.

WRITTEN BY