AI project at St. Mary School advancing to examine teaching ability of chatbots

Michael Tscholl
Michael Tscholl

Michael Tscholl’s long-term project at DeKalb’s St. Mary Catholic School has entered its next phase.

Last year’s study by Department of Educational Technology, Research and Assessment researchers focused on 50 students in sixth-, seventh- and eighth-grades to measure their comprehension of AI – artificial intelligence.

Part of that work lifted the curtain on how AI functions “under the hood” as well as how to recognize it.

Now Tscholl is ready to examine whether and how chatbots such as ChatGPT can effectively enable middle-schoolers to learn STEM (science, technology, engineering and mathematics) lessons as well as computer programming.

His ultimate goal: teaching children how to encode robots.

“Teaching children responsible, ethical use of chatbots and AI in general is critical to prepare them for the future,” says Tscholl, an instructor of educational technology.

Michael Tscholl guides a St. Mary student in programming movement commands for LEGO cars.
Michael Tscholl guides a St. Mary student in programming movement commands for LEGO cars.

“Our project is premised on the assumption that question-asking a chatbot is similar to a problem-solving process, in particular computational problem-solving and thinking. To obtain useful answers, question-asking involves problem analysis and problem framing, problem formulation, solution (answer) analysis, error correction and pattern recognition, etc.,” he says.

“In short, students will need to learn how to ask questions and critically analyze the chatbot’s answers,” he adds. “They will also need to discover patterns in question-answer interaction to understand how a chatbot works.”

Joining forces with ETRA research assistants Sahar Yaghoubzadeh Torki, Lida Niu and Md Ehtesham Adnan, Tscholl is leading the design of age-appropriate, complex-problem scenarios that require drawing on science knowledge and then involve developing Scratch programs such as science simulations.

“We aim to identify relationships between students’ question-asking and learning STEM and programming, and then examine the impact of helping students developing better questions and analyzing answers on student-chatbot interactions and on learning,” he says.

Good timing, he adds.

“Chatbots are increasingly used by students at all levels of education,” Tscholl says. “This use brings with it problems and challenges, including irresponsible use of chatbots, unwarranted trust in a chatbot’s answers, privacy issues, emotional connections and more.”