• Dr.A. Cineka

  Author

Automated Essay Scoring (AES) technology is rapidly increasing in teaching. Given AES technology's ability to scale and reduce the burden of grading, many institutions have adopted this technology. However, concerns have grown regarding AES and effects on minority and other marginalized groups. This essay reviews the small amount of available and relevant literature pertaining to Automated Essay Scoring (AES) to provide an understanding of the research regarding racial, linguistic, and social class biases. Existing literature suggests that, because of societal biases, AES undermines the integrity of assessment. The review also elucidates the primary concerns of AES, such as privacy, transparency, and the technology's self-proclaimed 'predictive' nature, and assesses a host of examples that employ different permutations of machine learning and deep learning. These systems demonstrate that bias in AES is still a concern. The review argues that there is an absolute necessity to offer an extensive array of creative data, provide clearer systems of scoring, and develop balanced and just frameworks to ensure fairness in Automated Essay Scoring systems. This review will spur more conversation about enhancing AES systems and moving toward a more transparent, accountable technology that improves the quality and integrity of educational assessments.

• PDF