AI-Driven Educational Assessments: Navigating the Intersection of Innovation, Equity, and Ethical Concerns for Future Learning

Abstract

The implementation of Artificial Intelligence (AI) to assess examinations and tests is transforming traditional evaluation methods employed by educational institutions. This advancement not only generates the employment opportunities outlined in the present study but also introduces technological progress that becomes more adaptable and offers personalised feedback to students. The development of cutting-edge technologies, although not yet widely accessible, undoubtedly opens new avenues of opportunity for subsequent generations. Given this context, this study conducted a systematic literature review (meta-analysis) of the current use of AI in educational evaluation, paying special attention to computer-assisted marking, personalised testing, and AI-based exam proctoring. Although AI technologies are expected to contribute to objectivity and standardisation, particularly in mass assessments, they also raise difficult-to-solve problems concerning assessment of advanced cognitive abilities, pitfalls regarding bias, and ethical considerations. While computer-based grading applications such as E-rater save time, their ability to assess creativity and sophisticated analysis is limited, often leading to a homogenised writing process. While adaptive testing tools, such as ALEKS (a web-based learning and assessment system), essentially perform some measure of customisation to the educational journey, they could also be somewhat limited in scope, restricting students to answer recall rather than more complex problem-solving tasks. However, language assessments using AI technology may result in an unfair bias, particularly in those with a non-native accent. Similarly, the use of AI tools for proctoring during remote examinations raises concerns regarding the invasion of personal privacy and overmonitoring. This study aims to address the implications of data-driven AI applications for educators and policymakers. It promotes hybrid processes that pair AI applications with human control to ensure fairness, transparency, and equity in academic evaluation.