Portal de Periódicos da CAPES

Exploring natural language processing in mechanical engineering education: Implications for academic integrity

2023; SAGE Publishing; Volume: 52; Issue: 1 Linguagem: Inglês

10.1177/03064190231166665

2050-4586

Jonathan Lesage, Robert W. Brennan, Sarah Elaine Eaton, Beatriz Moya Figueroa, Brenda McDermott, Jason Wiens, Kai Herrero,

Adversarial Robustness in Machine Learning

In this paper, the authors review extant natural language processing models in the context of undergraduate mechanical engineering education. These models have advanced to a stage where it has become increasingly more difficult to discern computer vs. human-produced material, and as a result, have understandably raised questions about their impact on academic integrity. As part of our review, we perform two sets of tests with OpenAI's natural language processing model (1) using GPT-3 to generate text for a mechanical engineering laboratory report and (2) using Codex to generate code for an automation and control systems laboratory. Our results show that natural language processing is a potentially powerful assistive technology for engineering students. However, it is a technology that must be used with care, given its potential to enable cheating and plagiarism behaviours given how the technology challenges traditional assessment practices and traditional notions of authorship.