Challenging Novice Programmer GenAI Usage During Code Comprehension Tasks with a Predict-Run-Investigate Protocol
Abstract
The introduction of generative AI (GenAI) has changed how novice programmers learn computer programming. Before GenAI, students learned to read, trace, and then write and debug programs. With the arrival of GenAI, students have formulated their own approaches to learning programming with the help of GenAI. This study seeks to explore and describe how students in an undergraduate, introductory programming course use GenAI tools when attempting to explain a series of Python code examples using a well-established code comprehension assessment technique known as explain-in-plain-English (EiPE). In addition, in an effort to challenge their understanding, they are asked to both explain the code and then predict and verify their understanding against a series of test cases by running the code. This challenge approach follows the predict, run, investigate, modify, and make (PRIMM) process, a Vygotskian-inspired instructional approach. By examining how students use GenAI and then providing them with a productive failure-inspired strategy, this work aims to both document how students are using GenAI to help learn programming without explicit instructional aid and how they adapt their approach when provided with a complementary approach.