Engineering instructor embraces AI as a teaching tool after summer course

 

Engineering instructor David Lawrence returned to Culver Academies with a new perspective on how students can use artificial intelligence responsibly and effectively after completing a seven-week summer course at the University of Wisconsin.

Lawrence previously had told his students they were not allowed to use AI. But he knew that because artificial intelligence and machine learning have become increasingly common in engineering and science research, he needed a deeper understanding of how the technology works before teaching students to use it.

"I really like the idea of getting my hands involved in that and then knowing that I want to be able to teach my students how to use AI and machine learning appropriately," Lawrence said.

Lawrence spends most of his summers taking such classes because engineering and technology continue to evolve rapidly and he believes as a teacher at Culver Academies, an elite leadership-oriented boarding school in northern Indiana, he has a responsibility to keep pace with those changes.

"It helps me also see what's going on engineering-wise at different universities," he said. "How are they teaching it? How are they addressing this issue? It helps me in preparing students for engineering at the college level because I now am seeing firsthand what is happening at that college."

For Lawrence, professional development also serves another purpose: sharing that excitement for discovery with students.

"I love to learn and of course that goes into my classroom because I'm excited about it," he said, "and then hopefully my students see that and it inspires them to do research."

One of the most eye-opening moments of this past summer came when Lawrence observed how university researchers use AI to generate computer code.

Rather than building every program from scratch, professors and researchers frequently used AI tools to create a foundation of code before modifying it for their specific projects.

"The professor right away said, ‘Let's go to Google AI and write this prompt and generate a base code for us and then we can work from there,’ " Lawrence said.

That experience challenged some of his assumptions about artificial intelligence and revealed a role for AI that goes beyond concerns about plagiarism or shortcuts.

"So just the use of AI in generating code and how well it does generating the code and provides you with the organization of the code and the comments throughout," he said. "If you have a basic understanding of programming, using AI to generate that code is much more efficient."

The key, Lawrence emphasized, is that students must still understand the underlying concepts.

 

 

AI can generate code, but users must know how to evaluate it, modify it and determine whether it is producing accurate results.

As a result, Lawrence plans to incorporate AI instruction into his classes.

"I've already started with my honors research classes and told them, the default is don't use it without my permission and instruction," he said.

He also intends to have students document the prompts they use and the responses AI generates, creating a record of how they interacted with the technology and how they improved upon the results.

The program helped Lawrence improve his own prompt-writing skills, which he said are essential for getting useful results from AI systems.

Lawrence believes those skills will help students conduct research more efficiently and prepare them for college, where AI is becoming increasingly integrated into coursework and research.

Lawrence said one reason he chose the Wisconsin program was its emphasis on AI and machine learning. While he understood the basic concepts, he wanted a better grasp of how machine learning systems are built and used.

The program introduced participants to the way machine-learning models are built, refined and improved.

"I knew the basics of machine learning, but what does that mean in a program?" he said. "I learned a little bit about that, but mostly about how to tweak that model in such a way in order for it to learn faster or become more accurate of a model."

Alongside the AI coursework, participants completed a Scholarship of Teaching and Learning project, which uses research methods to study student learning and classroom practices.

Lawrence plans to research how students learn programming concepts in engineering courses. His study will allow students to choose between written instructions, video instructions or both while learning programming skills. He will then track which method students use and how effectively they retain and apply the material later.

Lawrence also applied what he learned during the program to a research project involving northern Indiana frog populations.

Inspired by a summer science coaching workshop, he developed a machine-learning model designed to identify northern Indiana frog species from recorded calls.

Scientists often use frogs as biological indicators because healthy wetlands support diverse frog populations. Traditionally, evaluating frog populations requires researchers to listen to recordings and identify species.

Lawrence sought to automate part of that process. He gathered audio samples representing 16 frog species found in northern Indiana, locating at least 20 recordings for each species to train the model.

"So sitting in that workshop, it had me think of this idea of, ‘Hey, I can use AI in that,’ " he said.

The model currently identifies frog species from their calls. Lawrence hopes the project could eventually lead to tools that help researchers monitor wetlands more efficiently.

To test the model, he and his daughter collected recordings at a wetlands area near campus.

"I had to then learn some, I don't remember all 16, but about five or six different frogs," Lawrence said. "I'm like, ‘Oh, I know what that is now.’ " Lawrence said his goal is not simply to teach students how to use AI, but how to understand it.

As artificial intelligence becomes increasingly common in engineering and science, he hopes students will learn to use it as a tool while still developing the knowledge and critical-thinking skills needed to evaluate its results.