A Humanoid Social Robot as a Teaching Assistant in the Classroom

Thomas Sievers · 2025

Read A Humanoid Social Robot as a Teaching Assistant in the Classroom by Thomas Sievers (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-05646.txt.

Introduction

Thomas Sievers reports a classroom trial of a humanoid robot teacher's aide in this 2025 paper. Schools need relief for overburdened staff, yet social robots remain rare in lessons. The robot Pepper, linked to a large language model, presented new material to groups of high school students.

Sievers tested what the setup could do on site and surveyed students afterward. Every participant judged the robot's presentation of the material appropriate or at least partly so. Students also felt that using the robot made sense. The study offers an early, small-scale sign that embodied robot help can earn student acceptance in real classrooms.

Catalog record

CREATORThomas Sievers
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE21 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-05646

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Humanoid Social Robot as a Teaching Assistant in the Classroom Thomas Sievers Abstract Although innovation and the support of new technologies are much needed to ease the burden on the education system, social robots in schools to help teachers with educational tasks are rare. Child-Robot Interaction (CRI) could support teachers and add an embodied social component to modern multi-modal and multi-sensory learning environments already in use. The social robot Pepper, connected to the Large Language Model (LLM) ChatGPT, was used in a high school classroom to teach new learning content to groups of students. I tested the technical possibilities with the robot on site and asked the students about their acceptance and perceived usefulness of teaching with the help of a social robot. All participants felt that the robot’s presentation of the learning material was appropriate or at least partially appropriate and that its use made sense. 1 Introduction Although innovation and the support of technology are urgently needed to ease the burden on the education system, especially in today’s world, social robots are still rare in children’s school environments. But with Child–Robot Interaction (CRI) they can be effective and helpful tools that support children’s learning. They achieve better learning results than virtual agents and are comparable to human tutors if the tasks are simple and social [ 19 ] . Children learn more easily in multimodal and multi-sensory environments. Artificial Intelligence (AI) and Large Language Models (LLMs) can serve as tools for this purpose. However, AI systems developed for children are often based on individual activities on a screen and ignore an essential element of child development: interaction with a social environment. If children learn or solve a problem together, they retain what they have learned better and at the same time develop numerous other skills such as empathy, theory of mind, metacognition and emotion regulation [ 23 ] . There is also good experience with social robots that help children with autism spectrum disorders to learn while taking their special needs into account [ 26 , 16 ] . Language education, robotics education, teaching assistance, social skills development and special education as well as guided learning through feedback are named as the five most important applications of educational robots, and preschools and primary schools are seen as having a large potential for the use of educational robots in the near future [ 9 , 21 ] . Social robots are used as moderators in a collaborative learning process and can create a pleasant learning experience for learners [ 6 , 7 ] . Children’s valence and engagement in language learning can be improved through personalized tutoring from a robot, even when used in the natural environment of preschool classrooms during regular activities [ 12 ] . One major challenge when using social robots as intelligent learning systems in the classroom is to combine the robot-centered perspective, i.e. what robots are technically capable of, with the point of view of the child-centered perspective, which represents how the child or children can benefit from the robot and how the robot should act in order to best support them in achieving the interaction goals [ 17 ] . Even if a human who is criticized by a robot tends to have a negative reaction to the robot when it gives a poor evaluation, the evaluation by a robot instead of a human teacher could be in the robot’s favor because a robot tutor is free from bias toward the students’ gender, ethnicity, socioeconomic status, personal preferences, or other considerations [ 25 , 22 ] . In this paper, I summarize findings from testing the humanoid social robot Pepper in interaction with students at a high school in Germany. The robot’s main task was to support lessons by presenting new learning content in conversation with small groups of students. To make the robot speak I applied the LLM of OpenAI’s Generative Pretrained Transformer (GPT, commonly known as ChatGPT) [ 15 ] . I chose the approach of on-site practical testing for the question of the feasibility of using a humanoid social robot as a teacher’s assistant in a normal school classroom. 2 Related work There are many studies on the suitability of robots in education for children of different age groups with various focuses. Woo et al. investigated field studies on the use of social robots in real-life classrooms, which showed that the use of social robots in natural school environments is feasible, but that there are difficulties in personalizing the interaction, among other things [ 24 ] . Various types of breakdowns, including misunderstandings, malfunctions or inconsistencies in the interaction between children and a social robot tutor at an elementary school were identified by Serholt [ 18 ] . Other studies suggested that social robots could be used as language tutors for children from an early age [ 3 , 4 , 13 ] . This seems to be one way of compensating for language deficits in the school environment. Personalization and adaptation on the part of the tutor to the student’s current abilities appear to make sense. However, Gao et al. used a Pepper robot as a tutor to help people solve logic puzzles. Their results showed that the personalization of the robot’s behavior on user’s task performance had a negative influence on people’s perception. A more varied behavior of the robot was preferred [ 11 ] . They came to the conclusion that caution is required when developing social and adaptive behaviors in robots designed to support human learning. A review conducted by DiPaola et al. on social robots and children’s rights found that in Human-Robot Interaction (HRI), nonverbal interaction related to the physical nature of robots is considered as important as verbal interaction and that robots are differentiated from other AI-based devices by their embodiment [ 10 ] . Most of the papers examined in the review raised concerns about the physical safety of children, and only a few papers referred to the psychological or mental safety of children. While inclusion seems to be one of the most popular research topics in the field of children’s rights, other aspects such as explainability and fairness have not yet been sufficiently researched. Concerns that a robot tutor could have a negative impact on children’s psychological well-being and happiness, as well as trust and privacy, may be offset by evidence of great opportunities in terms of improving educational outcomes and freedom from bias [ 22 ] . Apart from particular approaches for inclusion, autism, language learning for migrants or similar special cases, my interest lies in the use of a social robot in the classroom at a typical school with ordinary students. 3 Methods Studies show that the acceptance of robots by children depends to a large extent on the age of the child and the characteristics of the robot [ 19 ] . In a real school environment, it is technically very difficult to enable a robot to understand conversations with people in background noise or in groups using today’s technology. This complicates the realization of real social interactions immensely [ 14 ] . However, the possibility of using ChatGPT to realize the robot’s communication capabilities has significantly reduced

Read the complete text