Developing Human-AI authoring systems for the creation of study and practice materials

Abstract

This internship was centered around the goal of developing a human-AI authoring system to facilitate the creation of study and practice materials. The main development involved with this project was associated with the creation of a software library that, when called, will prompt a generative AI to create questions of various formats related to the material provided. To accomplish this, the library is passed a text file with specific param- eters, like an integer to specify the number of questions that the user would like it to output. The end goal being to implement it into a pipeline that intakes an audio file and outputs these questions. Within the pipeline, the audio file is passed to an Automated Speech Recognition software. A transcript is produced which acts as the text file in the developed software library to create practice questions. To better visualize the output, an interactive web application was developed with Django. Output was also evaluated to rate the questions generated by the AI. Overall, output was judged to be acceptable for usage when paired with the review of an instructor.