Criticize my Code, not me: Using AI-generated Feedback in Computer Science Teaching

Authors

DOI:

https://doi.org/10.34190/icgr.7.1.2002

Keywords:

Female in STEM, Large Language Models, ChatGPT, Feedback Process, AI-based Feedback

Abstract

Large Language Models (LLMs) like ChatGPT can help teachers to tailor learning tasks for their students, combining learning objectives and storytelling to raise interest in the subject. AI-based learning task design can help to support competency-based learning, especially for girls in STEM courses like computer science, where otherwise the “Leaky STEM pipeline” (Speer 2023) leads to a constant loss of female students over school time. LLMs support many steps of the creation cycle of learning tasks. One important step is the feedback process between teachers and students during and after solving the tasks. Students need person-related as well as process-related feedback to make progress. Sometimes problems occur when teachers give feedback in a way that embarrasses or hurts the students. Especially female students often need more confirmation to make them aware of their progress, but studies show that boys demand and get more attention by teachers in this situation. This is one of the many reasons why girls lose motivation and interest in STEM courses over time. Since male and female teachers differ in expressing feedback without being aware of it, it is necessary to raise their consciousness. LLMs like ChatGPT can be used in two scenarios here. The first scenario is helping teachers to formulate objective feedback in a way that is adequate and understandable for the target group – e.g., young girls or boys - in a specific situation. The second scenario is training the teacher in a Socratic way, where the LLM simulates a student receiving the feedback and reacting to it according to established communication models like the Four Ears-model by Schulz von Thun (Schulz von Thun 1981) or Berne’s Transactional Analysis (Berne, 1964). This case study provides examples and prompting schemes for both scenarios and discusses the fragile balance between avoiding gender stereotypes in LLMs and giving more helpful and sustainable feedback for female students to foster self-esteem and competency-awareness.

Author Biographies

Sibylle Kunz, IU Internal University

Sibylle Kunz is a Professor of Media Computer Science at the IU International University of Applied Sciences. She earned a Diploma in Information Systems at Technische Universität (now Hochschule) Darmstadt and received her PhD in Digital Humanities from the Friedrich Alexander University Erlangen-Nürnberg. She also owns an IT consulting company.

Adrienne Steffen, IU Internationale Hochschule

Adrienne Steffen is a Professor of Business Administration at the IU International University of Applied Sciences. She earned a BBA in International Management, during which she spent time at the University of Michigan and at ESC Rennes. She received her PhD in Marketing from the University of Strathclyde. Adrienne worked in marketing and business development.  

Downloads

Published

2024-04-18