Teachers tend to repeatedly assist the same students when using artificial intelligence-powered educational tools, according to a study released on Apr. 7 by researchers at North Carolina State University and Carnegie Mellon University. The findings suggest that new software features could help educators track their interactions and ensure all students receive adequate attention.
The study addresses how teachers distribute their time among students in classrooms that use intelligent tutoring systems, or ITS, which are AI-driven programs designed to offer personalized feedback and monitor student performance. According to Qiao Jin, first author of the study and assistant professor of computer science at North Carolina State University, “AI-powered tools are increasingly common in K-12 classrooms, but teachers still play a critical role.” Jin said the research aimed to examine how teachers decide which students need help and how they allocate their time.
Researchers interviewed nine middle school math teachers who used ITS in their classes. These interviews revealed that while it would be ideal for teachers to spend one-on-one time with every student, this is not possible due to practical constraints. Teachers reported making decisions about whom to help based on several factors, especially whether a student had previously needed assistance and each student’s engagement state—such as struggling with problems or being idle.
Jin explained that “ITS can notify teachers when students have been consistently entering incorrect answers or have not interacted with the system for an extended time,” referring to these as ‘struggle’ and ‘idle’ states respectively. To observe these behaviors in practice, researchers analyzed over 1.4 million interactions from 339 students across 14 classes during the 2022-23 school year.
The data showed that “teachers are more likely to interact with students that they have interacted with before, even after considering who is engaged and disengaged in the classroom,” Jin said. She added: “Basically, if a teacher has intervened to help a student in the past, they are more likely to intervene to help that student in the future.” Jin also noted that each teacher brings personal definitions of fairness shaped by training and experience.
The paper will be presented at the Learning Analytics & Knowledge Conference (LAK26) later this month in Bergen, Norway. The authors believe these insights can inform new dashboard features for ITS platforms so educators can better balance attention among all pupils: “Teachers have a difficult job and developing better tools to help them do that job effectively is worthwhile,” Jin said.
