Education in transition : a study on artificial intelligences’ performance in university software engineering courses
Year: 2024
Type: dissertation
Abstract: Release of ChatGPT and the generative artificial intelligence (AI) boom following it has brought much interest and a large user base to AI software that can create text, images and videos from user input and answer any questions they have. As software like these can produce artefacts that look like human made, there is a possibility that students could easily skip manual work in education by using these AI tools. \n \nThis study aims to find out how well different generative AIs can pass exercises in university software engineering courses. First a literature review was conducted about what research has already been done on the topic of this study. After this, a survey was done to see what different generative AI software and models are available to consumers and a few of these were selected to measure their performance in weekly exercises from LUT university's software engineering courses. These exercises feature tasks in various programming languages and quizzes on course topics. \n \nResults show that selected AIs can pass most, if not all, exercises and can often achieve moderately high points in course grading. This study does not test AIs on other parts of courses like larger course projects and exams, or in other exercises than programming tasks and quizzes and thus does not show how well AI alone could pass whole courses or other types of exercises.
Source: LUTPub (LUT University)
Author: Suomalainen, Eero
Institution: Lappeenranta-Lahti University of Technology
Language: English
Subfield: Education
Field: Social Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W6981323755
