Improving learning outcomes with AI

AU researcher Dr. Ali Dewan explores ways to automate feedback to help students improve performance using artificial intelligence

It’s not always easy for instructors in an online learning environment to know how well students are engaging with and understanding the course material… yet. 

Athabasca University researcher Dr. Ali Dewan, an associate professor in the School of Computing and Information Systems in the Faculty of Science and Technology, is developing an artificial-intelligence-based application that would be able to provide that kind of feedback to students or their instructors instantly. 

“In online learning, the students are isolated from their instructors. It’s not like in person,” he said. “I’m trying to identify how much the students are engaged with their learning activities, especially like when they watch a video lecture or when they go through any learning material.” 

This work, which is funded through a Natural Sciences and Engineering Research Council of Canada Discovery Grant, has two main elements. The first is facial recognition to help gauge students’ emotional and behavioral engagement as they work through material. The second is textual analysis of a student's written submissions to help understand their level of cognitive engagement.

After detecting what types of challenge the student is facing, identifying those challenges and providing adequate feedback to students is the next step.

Dr. Ali Dewan, professor, Faculty of Science and Technology

Detecting student engagement 

With in-person learning, an instructor can look around the room to see which students are engaged with the lecture via facial expressions and body language.  

This is not currently the case with most online learning platforms, including what’s used at AU. Instead, students engage with the material on their own. Often, the only way instructors can measure how well a student is learning the material is by marking assignments and exams. 

By using facial recognition technology, Dewan hopes to develop an AI application that would be able to gauge a student’s level of interest in and engagement with any given piece of material by measuring and analyzing the student's physical cues. He hopes this kind of analysis could not only help identify that a student is disengaged, but could also identify the reason for that disengagement—whether it’s poorly designed course material, the student is just distracted, or even that they are missing important background information. 

This same research involves creating an AI application that could review a student’s post to a course’s discussion forum to determine how well a student has understood the course material and achieved the learning outcomes. 

“This is at the preliminary stage where we are just trying to detect it,” he said. “After detecting what types of challenge the student is facing, identifying those challenges and providing adequate feedback to students is the next step.”

Ethical challenges with AI applications

While this research has great potential to support teaching and learning, Dewan said there are ethical considerations guiding this work. 

Specifically, a student would need to consent to having facial recognition technology used while they do their coursework. As such, this research is currently based on open data sets that are publicly available. 

The ethical considerations also help drive the thinking behind how to implement this kind of application. Dewan said specific information about emotional engagement, like how many times a student yawns or scratches their ear, might not be something they’d want to share with an instructor. As such, he envisions that this component of the application would exist locally on a student’s personal computer, rather than being something that would run through a university’s online learning platform. 

“The information would not be shared with others,” he said. “It would only be shared with students, so in this case students might be more comfortable using this kind of system.” 

This limitation would not necessarily affect analysis of forum responses, though, as those posts are already shared with classmates and the instructor. 

“It really depends what kind of information we’re analyzing to get the final result,” Dewan said. “We’re trying different modalities in this case.” 

This work has improved my AI knowledge, analytical skills, and shown me how technology can profoundly shape the future of education.

Dharamjit Parmar, MScIS student and research assistant

AI projects to explore

While detecting student engagement is Dewan’s main research focus at this point, he is also engaged in several other projects, including collaborations with students and other researchers. 

One of his graduate students is working to develop an educational dashboard that would allow students to see how their own performance ranks when compared to aggregate data from their classmates. This could help students improve their time management, understand areas where they may be falling behind, and potentially how well engaged a student is compared to their classmates. 

“This dashboard is very student focused,” he said. “The information is not shared with an instructor, rather a student can monitor their progress in a course within the dashboard.” 

Another project he’s working on, in collaboration with another student, is creating an AI application that could automatically score a student’s posts to a class forum based on a rubric the instructor provides. Automating the grading process for this type of assignment could save time for instructors, and also allow students to get instant feedback. 

While Dewan benefits from working with student researchers, the students themselves also get a lot out of the process. Dharamjit Parmar, a research assistant and student in the MScIS program, said working with Dewan to study cognitive engagement detection with AI has been a defining part of his educational experience. 

“It is not just about advancing technology, it is about understanding how people learn and building tools that make education more adaptive and inclusive,” he said. “This work has improved my AI knowledge, analytical skills, and shown me how technology can profoundly shape the future of education.”