Creating test questions that are valid and reliable usually takes time, expert review, and costly testing. This is especially challenging for large-scale educational assessments where every item must meet high standards.
This invention uses AI and psychometric feedback to automate the process. Starting with a small set of expert-written questions, the system evaluates quality, trains a language model to generate new items, and filters results through psychometric analysis. This cycle repeats, improving item quality over time.
The system can run with or without human oversight, making it flexible for different use cases. It helps education companies and assessment providers scale faster, cut development costs, and maintain rigorous test quality.
The image is a flowchart showing how expert-reviewed test items are sorted, used to guide AI generation, and filtered through psychometric analysis to build a stronger question bank over time.