AI-Automated Assignment Scoring to Scale a Professional Development Micro-Credential Program
International Journal on E-Learning · 2024 · 24 (3) · 335-345
DOI: 10.70725/881380vlsnpa ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal on E-Learning, 2024. Verfasst von Cathy Cavanaugh und 2 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref).
Abstract (Originalsprache)
To address needs in one US state to provide a professional development micro-credential for tens of thousands of educators, we automated an assignment scoring workflow in an online course by developing and refining an AI model to scan submitted assignments and score them against a rubric. This article outlines the AI model development process and pilot outcomes. The AI system employs a combination of pre-built machine learning models and custom reinforcement learning algorithms to evaluate and enhance learners' capabilities through a continuous active feedback loop. Trials of the AI assignment scoring system in the course showed that the model became more accurate, learner mastery rates in courses using the system were equivalent to or better on average compared to mastery rates in courses not using the system, and instructor time saved in scoring can reduce the cost of scoring by one third.
Zitieren
APA
Cavanaugh, C., Humphrey, B., & Pullen, P. (2024). AI-Automated Assignment Scoring to Scale a Professional Development Micro-Credential Program. International Journal on E-Learning, 24(3), 335-345. https://doi.org/10.70725/881380vlsnpa
Vancouver
Cavanaugh C, Humphrey B, Pullen P. AI-Automated Assignment Scoring to Scale a Professional Development Micro-Credential Program. International Journal on E-Learning. 2024;24(3):335-345. doi:10.70725/881380vlsnpa
BibTeX
@article{Cavanaugh2024,
author = {Cavanaugh, Cathy and Humphrey, Bryn and Pullen, Paige},
title = {AI-Automated Assignment Scoring to Scale a Professional Development Micro-Credential Program},
journal = {International Journal on E-Learning},
year = {2024},
volume = {24},
number = {3},
pages = {335--345},
doi = {10.70725/881380vlsnpa}
}
EUVIMED