The Texas Education Agency (TEA) implemented an automated scoring system utilizing artificial intelligence to grade openended STAAR writing and essay responses for more recent testing in schools that have spiked zero scores.
I STAAR AI scoring being questions by school districtsn 2026, STAAR tests were administered primarily online and graded using a combination of automated computer scoring for openended written answers and standard scale scoring for the overall exams.
This hybrid scoring method assigns 75% of grading to AI and 25% to human graders. The controversial change has prompted many school districts and parents to request manual rescores.
The TEA is defending its decision to let computers grade students’ STAAR essay test questions in the face of mounting criticism from parents and teachers who question the fairness of an automated scoring system.
The grading and evaluation process involved AI-assisted automated scoring engine, powered by natural language processing, to evaluate the increased number of open-ended, short-answer, and constructed-response questions. This technology assessed the written answers based on rubrics, grammar, and content accuracy. The AI system has resulted in a drastically higher number of students receiving zero on open-ended written responses compared to traditional human grading. Educators are concerned that the AI struggles to recognize creativity, unique writing styles, or responses containing colloquialisms and Spanish phrases, which can unfairly penalize students. According to regional Education Service Centers, if a student’s written responses were scored inaccurately by the computer, parents and educators have the option to request a manual human rescore.
For grades 3-8, these tests have no consequences on academic progress or grade promotion, however, if the student is taking highstakes End-of-Course (EOC) tests required for graduation, a successfully appealed score will be updated Teacher unions, lawmakers and parents say they were blindsided by the changes they fear could punish students who give creative or atypical answers computers don’t recognize. STAAR test results are particularly sensitive because they are used to determine the state’s A-F ratings for public schools.
Because STAAR scores directly dictate the A-F accountability system for districts, erroneous AI grading may lower campus ratings in several areas. If a campus receives five consecutive failing grades from the state, the entire school district could be placed under a state conservatorship.