Appropriations of Artificial Intelligence for writing by visually impaired people at the university: uses, barriers and risks
Visual impairment; Generative Artificial Intelligence; Writing; ChatGPT; Higher Education.
The objective of this qualification project is to examine the different ways in which Generative Artificial Intelligence (AI) has been appropriated by people with visual impairments and to critically understand whether AI for writing has been used by visually impaired students in conjunction with other technologies, and what these students say about its use, barriers, and potential risks in the context of generative AI appropriation for academic writing. The selected methodology included Content Analysis by BARDIN (1977) along with Data Triangulation by TRIVIÑOS (1987). Data collection involved semi-structured interviews with nine students, five from undergraduate and four from graduate programs at the University of Brasília (UnB), all with visual impairments. The participants included two people with blindness, two with monocular vision, and five with low vision. The theoretical framework and concepts about AI were based on authors such as KAUFMAN, 2018; BOA SORTE et al., 2021; COZMAN, PLONSKI, NERI, 2021; MEDEIROS, 2018; ROE; PERKINS, 2022; ROGERSON; MCCARTHY, 2017, in addition to the concept of appropriation as sense-making defined by BATISTA (2018). For this research, six databases were explored in search of academic precedents, but no results were found. Given these perspectives, we can understand that visually impaired people in the context of higher education still do not consider generative AI for writing to be a reliable and valid option, as reported experiences included access difficulties and distrust in the results.