CS professor Abdussalam Alawini receives SIIP award to integrate AI into engineering education

Siebel School of Computing and Data Science professor Abdussalam Alawini is the principal investigator for a new initiative under the Illinois Grainger College of Engineering Strategic Instructional Innovations Program (SIIP). SIIP is a competitive awards program designed to support innovation in education, similar to the research model at the University of Illinois Urbana-Champaign – “an engaged community, collaborative projects, faculty-led innovation, rigorous evaluation, and a scholarly approach to pedagogical methods.”

The project is dedicated to integrating AI into engineering education, primarily focusing on studying its potential impact and benefits.  The cross-disciplinary team consists of Alawini with co-investigators  Volodymyr Kindratenko (NCSA, ECE), Sotiria Koloutsou-Vakakis (CEE), Tomasz Kozlowski (NPRE), Christopher Tessum (CEE), Meredith Blumthal (ACES) with Maryalice Wu (CITL). Their proposal,  “Exploring the impact and potential of Generative AI in Engineering Education,” aims to integrate generative AI into STEM education, ultimately enabling students to “examine problems, investigate solutions from different perspectives, and identify solutions that are not bound by traditional resources or methods. To help students make connections of the information they learned, we aim to integrate their personal discoveries with broader knowledge through the use of AI."

Photo Credit: University of Illinois / Holly Birch Photography

Alawini says the team began to form when he connected with Kindratenko. “I've been interested in using AI in education before this emergence of generative AI tools,” he says. “I've been giving workshops to staff at the Business Office, to James Scholar students, to students here at U. of I., and I know that professor Kindratenko and all kinds of folks are trying to develop technologies. I heard he's part of this Amazon funding to deploy AI domain experts.  I heard about this as part of the Education Innovation Fellowship. So, I reached out, and we started brainstorming ideas and collecting the people interested in this idea.”

The Kern Family Foundation is a second source of project funding, and the team is aligning with the KEEN framework to build survey tools to determine how engineering instructors and students currently use generative AI tools.

Kindratenko and Alawini have experience in utilizing generative AI in education. Kindratenko designed the UIUC AI Chatbot platform to aid students’ questions about videos, text, slides, and other course materials. It has been used in courses such as ECE 120, ECE 220, ECE 408, NPRE 247, and NPRE 598. The team proposes to expand the platform’s application to other STEM disciplines via improved integration of GPT-4 with customized AI agents.

Alawini says he was aware of similar efforts at other units in the Grainger College of Engineering. “I heard a lot of conversation from other teams, in particular, the teams trying to introduce computing into civil and environmental engineering. We had a lot of questions about what we are going to use, what we are going to do with AI, and how we are going to introduce computing. Are we going to need computing to begin with for our classes, given the emergence of these generative AI tools?  I've seen the need, and it was easy for me to reach out to people within that team to say, ‘Hey, we're working on this project. Are you interested in not just trying to think about what's going to happen, but be part of it and try to help us figure out the landscape of AI in engineering education?’”

Alawini has focused on the challenges students face when learning database programming. In their proposal, the team plans on using the “advanced capabilities of a Generative Pre-Trained Transformer (GPT) model with an online assessment feedback system… to enhance both the accuracy and effectiveness of error detection and feedback.” A diverse group of student submissions were used to “fine-tune” the GPT models. The models became better at recognizing programming errors and thus giving students constructive suggestions for programming.

He explains, “We use PrairieLearn as an online management assessment tool so students can do their homework, in-class group activities, homework, quizzes, and exams on PrairieLearn, and that captures all the data, like any submission to a particular problem will be captured with other information. We've noticed that students struggle with these types of problems.” Alawini says, “Students may encounter two different types of problems. They may encounter a sentence problem where they wrote a query that the language could not compile. So, it says, ‘I don't understand you,’ that's, to some extent, easier to deal with because the compiler will tell you the problem. The most difficult ones are the semantic errors, where your program executes but does not pass the test cases. They're not getting the correct answer. Students spend a lot of time trying to submit and resubmit, and optimizing. Getting AI as a tutor when we see students getting to the point of frustration. We've been custom-training fine-tuned models to act like an instructor or a TA to help guide the students when they need help.”

By analyzing data collected from student interactions, the team aims to gain insights into how these technologies affect learning outcomes, student engagement, and course structure. Such information will guide the development of best practices for incorporating GenAI into teaching strategies, ensuring that coursework is structured to maximize the benefits of AI while mitigating potential risks such as dependency or misuse. Policies will be designed to protect students and faculty. An environment can be fostered where GenAI tools are used responsibly, enhancing the educational experience without compromising academic integrity.

The team’s proposal states its ultimate purpose is “to catalyze a culture shift towards more dynamic and technology-driven teaching methodologies.” Outreach activities like workshops, seminars, and collaborative forums will build networks educators across diverse institutions who are interested in adopting GenAI into their curricula.

For the present, Alawini says, “We'll be focusing on the qualitative part. We're trying to run surveys and understand what's happening. We see students use GenAI in class.
I run collaborative learning activities in class and walk around the classroom to see what my students are doing. Some use these tools, and of course, we have different policies regarding that. We want to understand and influence policies and how we should restrict or encourage the use of generative AI tools. We believe that computing is involved in most engineering curricula. If we can build tools to help people learn within their classrooms, we can accelerate how computing helps within these disciplines, whether engineering or beyond. The focus right now is on engineering just for the scope. And then beyond that, we'll be certain that what we're building will have an impact.”

Grainger Engineering Affiliations

Abdussalam Alawini is an Illinois Grainger Engineering professor of computer science and is affiliated with the Siebel School of Computing and Data Science.

Volodymyr Kindratenko is Assistant Director at NCSA, an Illinois Grainger Engineering professor of electrical and computer engineering and is affiliated with Siebel School of Computing and Data Science. He is Director of the Center for Artificial Intelligence Innovation (CAII) at NCSA.

Sotiria Koloutsou-Vakakis is an Illinois Grainger Engineering professor of civil and environmental engineering.

Tomasz Kozlowski is an Illinois Grainger Engineering professor of nuclear, plasma and radiological engineering and is affiliated with Computational Science and Engineering program.

Christopher Tessum is an Illinois Grainger Engineering professor of civil and environmental engineering.