Student teachers face a common challenge in their academic careers – they have limited opportunities to interact with elementary school children prior to their final year of college. To fix this, a Kennesaw State University researcher is leveraging artificial intelligence (AI) to give education students lifelike interactions with a virtual student earlier in their training.

Through the support of a three-year, $300,000 National Science Foundation grant, associate professor of instructional technology Dabae Lee will explore how artificial intelligence can help simulate children in an elementary school classroom. Her research involves the use of a chatbot, a computer program designed to simulate a text conversation with human users.

The chatbot is designed to be another tool for student teachers to use in the challenging task of developing responsive teaching skills, the ability to respond to children’s unique ideas or ask questions to help them extend their thinking.

“A majority of pre-service teachers usually have no opportunity to practice responsive teaching skills in a teacher education program before being placed in field experiences and interacting with real children,” said Lee, who teaches in KSU’s Bagwell College of Education. “When they are given practice opportunities before student teaching, such opportunities are rarely authentic or effective.”

In some cases, a trained adult or instructor plays the role of a child for the entire class, during which only a few student teachers get to pose their questions. Student teachers can also role-play with their peers, which still fails to mimic an authentic situation.

While the chatbot Lee is developing is not a student, it would work in tandem with other classroom simulation technologies to provide a realistic experience prior to student teaching in a real school.

“What we were thinking is, how can we create an AI system in which the learner plays an active role,” Lee said. “But then we need to also think about what makes meaningful interaction between teacher and learner.”

During the development of the chatbot, student teachers can ask a variety of questions, and the bot responds to those questions. When they (student teachers) ask questions that the virtual student doesn't yet understand, the virtual student answers the questions based on her current understanding, revealing her misconceptions.

Lee said, "We don't view misconceptions as errors to be replaced or corrected, but rather as valuable assets that learners can utilize to refine, reorganize, and expand their knowledge."

These interactions provide the student teacher the opportunity to explore and extend the virtual student's thinking. According to Lee, the research team aims to assist student teachers in developing responsive teaching skills that enable them to leverage children's misconceptions as unique assets for building and expanding learners' knowledge.

Lee and her colleagues went through several iterations of development of a prototype of chatbot. Each iteration, the chatbot has gotten better at answering questions, allowing the student teachers to improve their responsive teaching skills, by responding to and interacting with the virtual student. Based on the evaluation from each iteration, several features were added, including showing the feelings of the virtual student with emojis and the student being able to learn from interactions with key questions. Several other chatbots are now in development.

“I am encouraged by the progress we’ve seen in the early iterations of the chatbot, and I look forward to expanding its capabilities in the near future,” Lee said. “Through this unique application of artificial intelligence, I hope that we generate new learning opportunities for our students and equip them with experiences that bolster their teaching capabilities once they enter the field.”