https://msucerl.org/cmse101/tags/grading/Recent content in Grading on AI and Society CourseHugoen-usThu, 21 May 2026 00:00:00 +0000https://msucerl.org/cmse101/use-cases/1-1-automated-essay-scoring/Thu, 21 May 2026 00:00:00 +0000https://msucerl.org/cmse101/use-cases/1-1-automated-essay-scoring/<h1 id="automated-essay-scoring-aes">Automated Essay Scoring (AES)</h1> <h2 id="context--systems-architecture">Context &amp; Systems Architecture</h2> <p>Automated Essay Scoring (AES) systems utilize natural language processing (NLP) pipelines—ranging from traditional feature-extraction regression models (such as ETS&rsquo;s <em>e-rater</em>) to fine-tuned transformer networks (such as BERT variants)—to evaluate student writing. These systems are widely deployed by major assessment corporations, including Pearson, ETS, and Turnitin, to grade high-volume standardized tests, admissions essays, and placement exams.</p> <p>In a significant recent escalation of this technology, the Texas Education Agency deployed an automated scoring engine in 2024 to grade the State of Texas Assessments of Academic Readiness (STAAR) exams. The agency projected that the system would save <strong>$15–$20 million annually</strong> by reducing the number of human temporary graders from roughly 6,000 to under 2,000. However, the rollout resulted in widespread school district complaints after a dramatic surge in zero-scores for responses that did not fit the model&rsquo;s expected syntax paths.</p>