PULLMAN, WA – Washington State University researchers have received a National Science Foundation grant to develop a virtual teaching assistant to help students in engineering classes.
As part of the $400,000, three-year project, the researchers will develop their own artificial intelligence bot that will be trained in helping students in engineering fields. They will test and evaluate the virtual teaching assistant in upper-level thermal fluids courses with student volunteers.
“We’re trying to figure out effective ways to use AI tools — because students are already using this,” said Jonathan Steffens, assistant professor in the School of Mechanical and Materials Engineering who is leading the project.
The thermal fluids courses are good examples of where students often particularly struggle with complex concepts and really need and seek tutoring support. Some of those students are already using AI effectively to help them learn concepts and material, but others aren’t.
“It’s kind of like the Wild West of how we incorporate AI into our pedagogy,” said Steffens. “I think engineering educators should be on the forefront of that intersection of good pedagogical methods and recent technology developments with AI. We want to leverage what’s good about AI and get our students to use it in an effective and prudent manner.”
It’s kind of like the Wild West of how we incorporate AI into our pedagogy. I think engineering educators should be on the forefront of that intersection of good pedagogical methods and recent technology developments with AI.
Jonathon Steffens, assistant professor
Washington State University
As part of the project, the researchers will assess the accuracy of the virtual teaching assistants’ responses, measure how well the student volunteers in the study understand the course material, and analyze conversation logs between the students and the teaching assistant. They also hope the bot will be able to give the professor feedback and suggestions on what the students are having questions about.
The project aims to combine state-of-the-art in AI language models with effective pedagogical techniques, such as inquiry-based learning and the Socratic method, said Steffens. To develop their AI agent, the researchers will work with existing, open-source large language models and will train them with question-and-answer data sets as well as with information about engineering concepts. They aim to integrate the bot into the university’s Canvas learning management system, making it easily accessible for the students.
“It’s going to be a fine-tuning process to create a basic framework,” said Steffens. “It’s going to hopefully act more like a human teaching assistant or professor would, by engaging in actual pedagogical methods, so the hope is, once this is all done, that we can basically publicly host this tool with a variety of engineering subjects built in.”
Could the project be a scary indicator that bots are coming for future jobs or that students are going to be taught by computers? Steffens said that, in fact, what’s currently happening is scarier.
“I’m worried that there is a growing divide between students who can effectively engage with AI and use it as a tool to assist them in doing projects and to learn and augment their understanding versus students who are just using it as a crutch — I do feel like there is potential for that kind of gap to widen,” he said. “The scary thing is doing nothing — this project is trying to counteract that.”
In addition to Steffens, faculty members on the project include Narasimha Boddeti, Berry Assistant Professor from the School of Mechanical and Materials Engineering and Yan Yan from the School of Electrical Engineering and Computer Science.
