Researchers at the University of Chicago, working with Chicago Public Schools and Chapin Hall, have tested the use of robot tutors in elementary school classrooms to examine how different communication styles affect students’ social-emotional learning outcomes.
The study, led by doctoral student Lauren Wright and supervised by Sarah Sebo, focused on whether robots need to simulate human-like personalities to be effective in teaching skills such as empathy, conflict resolution and problem-solving. The research was presented at the ACM/IEEE International Conference on Human-Robot Interaction, where it received the Best Paper Award.
The project was developed through collaboration with educators and administrators in Chicago schools. Researchers first observed classroom instruction and consulted teachers to understand how social-emotional learning is currently delivered. In many cases, these lessons take place in group settings, limiting opportunities for individualized instruction and reducing engagement among some students.
To address this gap, the research team designed an experiment involving 52 fourth-grade students. Participants were divided into three groups: one group interacted with robots programmed with fictional, emotion-based dialogue; another group worked with robots that communicated in factual terms and explicitly stated they did not possess feelings or personal experiences; and a control group continued with standard classroom instruction without robot support.
The robots adapted lesson plans from the Second Step curriculum into one-on-one interactions, allowing teachers to continue managing the broader classroom. According to the findings, students in both robot-assisted groups demonstrated improved understanding of social-emotional concepts compared with those receiving only traditional instruction.
The study also found that students working with factual, non-anthropomorphic robots often showed stronger engagement with lesson vocabulary and problem-solving language. Researchers indicated that fictional narratives and simulated emotions, commonly used in educational technology, may introduce distractions or reduce student comfort in some contexts.
Wright stated that the results challenge assumptions about the role of human-like traits in educational robots, noting that commonly used design approaches do not necessarily lead to improved learning outcomes. Sebo emphasized that the research is intended to support, rather than replace, teachers by providing tools that expand their capacity to offer individualized attention.
The findings come amid broader concerns about children forming attachments to artificial intelligence systems. By demonstrating that robots can be effective without mimicking human emotions, the study suggests an alternative approach that may reduce such risks while maintaining educational benefits.
The researchers conclude that robot tutors can function as supplementary tools in classrooms, particularly in areas where individualized instruction is limited, while preserving the central role of human educators.
