Abstract for: Threat of Generative AI on Academic Integrity and Possible Policy Implementations

Generative AI is a rapidly growing technology that has come to directly influence our lives. Most recently, the industry has been disrupted by OpenAI’s release of ChatGPT, a chatbot that can produce 5-page essays or complex Java programs within seconds. The technology has especially stirred the college population about cheating implications, leading to intense discussions on what actions to take to ensure academic integrity (McDade). This study uses a system dynamics approach to understand how the rising adoption of generative AI might influence academic dishonesty in the US college population over time and what some leverage points are for various policy options by building a model in the STELLA software. A tipping point dynamic is reproduced with three different adoption scenarios (slow adoption, baseline, fast adoption), all of which are tested with various policy options. The results show that intervention strategies that directly target the process by which students start to cheat are the most feasible in minimizing cheating, whereas focusing on the tipping point dynamic and catching cheaters limit the number of generative AI users overall. These findings can become a useful input for policy discussions among college administrators as an addition to learning and mental health considerations.