The US state of Texas Education Department will implement natural language processing systems in the upcoming weeks to evaluate students' responses in standardized tests. The department stands to potentially save at least $15 million annually.
As per The Texas Tribune's editorial team, the underlying machine learning technology is akin to GPT technology developed by companies like OpenAI, utilizing natural language systems. The system was trained using 3000 responses to open-ended questions from the State of Texas Assessment of Academic Readiness test series along with corresponding human evaluations.
The AI system will utilize this data to learn the characteristics of provided answers and assign scores similar to those of a human grader. In instances where the system is uncertain about an assessment, the answer to a question will be automatically referred to a human grader. The same applies if the system is unable to interpret an answer at a certain juncture. Additionally, reports indicate that parents and students have the option to request a reassessment of the tests if they disagree with the results, a process that will incur a $50 fee.
The Texas Education Department stands to realize cost savings between $15 and $20 million annually through this initiative. These savings are achievable as the department will require significantly fewer staff members due to the implementation of the AI system. In the previous year, the department employed 6000 temporary graders, while this year, fewer than 2000 workers will be needed.
English (US)