Multidimensional Research Insight is an international scholarly journal of social sciences which provides wide, excellent cross-disciplinary research papers. It aims at increasing the generation of new self-integration knowledge with an emphasis on interdisciplinary research which harnesses interests cut across discipline to find solutions to local and global amplifications. The published journal is intended to enhance the probability of domain spanning and allow researchers to focus on the discovery of the connections on one field and others, and provide the overviews which are not partial.
Article Views: 646
Background: The rapid integration of AI (AI) into educational assessment systems raised concerns regarding fairness, equity, and algorithmic bias, particularly for vulnerable student populations. Purpose: This study investigated equity and bias in AI-based educational assessments within the United Arab Emirates (UAE), focusing on SEND learners, gifted students, and gender-diverse groups. Methods: A convergent parallel mixed-methods design employed purposive sampling from 15 UAE schools. Participants included 400 students (ages 11-18 years, M=14.3, SD=2.31; 192 males, 185 females, 20 non-binary), 82 teachers (56 females, 24 males), and 28 school leaders. Quantitative data were analyzed using SPSS (t-tests, ANOVA, regression, Cohen’s d). Qualitative data underwent thematic analysis (Cohen’s kappa=.84). Results: SEND students experienced severe disadvantage across equity dimensions (d=0.76-1.12). Gifted students rated pedagogical value significantly lower (d=0.61). Female students perceived AI assessment as less fair than males (d=0.45). Teacher trust correlated exceptionally with transparency (r=.970, R²=.941). Hybrid human-AI models were universally preferred (M=4.25). Conclusion: AI assessments introduced systematic biases disadvantaging marginalized learners. Achieving equity required fairness auditing, transparent algorithms, inclusive datasets, robust governance, and preserved human judgment.
AI, Educational Assessment, Algorithmic Bias, SEND Learners, Gender Equity, UAE Education