Beyond the algorithm

A longitudinal study of academic integrity trends in the context of artificial intelligence in STEM education

Abstract

The rise of generative artificial intelligence (GenAI) raises critical concerns about academic integrity and the authenticity of student learning. This paper draws on academic integrity records from the South Australian Institute of Business and Technology to explore trends in GenAI use, detection, and institutional response following the public release of ChatGPT. Using a quantitative data analysis approach, we examined patterns across multiple study periods and disciplines to highlight key differences in how GenAI is being addressed in STEM and non-STEM courses. The findings reveal a rising number of misconduct cases, especially in essay-style and computer programming assessments. These shifts raise concerns about the effectiveness of current assessment models, and the long-term credibility of STEM qualifications. The paper argues for a proactive and pedagogically informed response, emphasising authentic in-class assessments, policy reform, and inclusive learning environments, to ensure STEM graduates are equipped not only with technical skills but also with the ethical and analytical capacities required for work in industry.