Understanding Generative Artificial Intelligence (Gen AI) as a Lab Partner: A Case Study in Engineering Education

Vollständiger Abstract

Worum geht es in dieser Arbeit?

To investigate the potential of generative artificial intelligence (Gen AI) in sustainable engineering education and laboratory instruction, this study used the direct shear test in soil mechanics as a case study. A total of 112 third-year undergraduate students majoring in hydraulic engineering were assigned to traditional and AI-assisted groups. Their performance in experimental operation, data processing, report writing, presentation of results and problem solving was compared using grade statistics, classroom observations and interview data. The results showed that the AI group generally outperformed the traditional group in Experimental operation, data analysis, report completeness, presentation and defense, and overall scores. The analysis showed that Gen AI can function as a form of cognitive scaffolding in conceptual explanation, data processing, report structuring, and error analysis. Specifically, it can provide stage-specific prompts for understanding, procedural support, and feedback during the learning process, thereby helping students complete experimental learning tasks. However, the qualitative corpus included documented cases of overreliance on Gen AI, including uncritical acceptance of generated outputs and alteration of discrepant data without adequate verification; these behaviours were not tabulated at the pair level, so their prevalence cannot be estimated. These findings suggest that the integration of Gen AI into soil mechanics laboratory instruction should be grounded in teacher guidance, disciplinary knowledge support, data verification awareness, and standardized tool-use practices. These findings also offer implications for the responsible use of Gen AI in sustainable engineering education and for the development of students’ AI literacy, awareness of data authenticity, and responsible technology use competencies.