Special Session--Scaling Automated Scoring: Addressing Practical and Conceptual Challenges

摘要

While automated scoring of constructed responses often makes it onto lists of how artificial intelligence (AI) technologies will change education, and constructed response scoring engines have attracted extensive research and development in the past few decades, adoption of these natural language processing and machine learning-based engines at scale has been slower than anticipated and desired, with significant public concern about the veracity of these solutions. This presentation will share a current real-world example to explicate both the benefits and challenges of scaling an automated scoring solution, a demonstration of an automated scoring engine including its key features and methods, and finally discussion of additional research required to further support adoption for a large population and diverse corpus.