Perception and Detection of Cyberbullying: A Comparison between the Educative Community and GenAI

Abstract

Bullying involves repeated and aggressive behaviors to intimidate or harm others, with cyberbullying being the kind that benefits from digital platforms to direct such messages. They both have serious consequences on students’ mental health and academic performance, demanding prevention and intervention strategies, including teacher training and the use of tools to detect these behaviors in university environments. Artificial Intelligence (AI), specifically generative AI, can automatically detect offensive language on digital platforms, resulting in an effective tool combating cyberbullying. This study investigates the capacity of members within a university community—students, faculty, and administrative staff—to perceive and detect cyberbullying in social media messages. The research utilizes generative AI tools to assess their effectiveness in recognizing cyberbullying patterns, comparing their results against expert evaluations. Results indicate that faculty members are most effective in identifying cyberbullying, while students show greater leniency, highlighting the need for targeted educational interventions. The generative AI models, despite limitations, demonstrate potential for early cyberbullying detection. Findings underscore the importance of training within educational communities and suggest that AI tools, when integrated into preventive programs, can enhance early intervention and promote safer digital environments.