Automated essay scoring (AES) is a technology that uses an algorithm to score student essays. The algorithm can be a set of rules, scoring rubrics, and machine learning models that can be applied to essays to determine their quality. AES is being used more and more among education technology platforms. A recent survey by The Learning Agency showed that there is a need for AI-based writing feedback tools: https://www.the-learning-agency-lab.com/the-learning-curve/teacher-digital-feedback-tools-can-help-reduce-grading-and-improve-instruction/
The Pros
AES is seen as a more consistent way to evaluate student essays because the computer cannot make subjective decisions based on race, gender, or socioeconomic status. To ensure that AES is not biased, researchers who build AES systems ensure that the training data for the AES system do not contain any biases. One can argue that AES is a more objective method of grading.
AES is also a more time-efficient method of grading essays, and we can use it to score a significant number of essays in a short amount of time. The computer can quickly score multiple essays, which frees teachers to focus on other tasks. Teachers spend up to 90 minutes a night grading papers, so automation can help with reducing teacher stress.
The Cons
AES assessments can be costly and time-consuming to develop, and they may not provide authentic, accurate, or fair measures of student learning. Because of the way AES works, it may not detect a student’s true potential. The algorithm behind the system is typically based on a scoring rubric or a set of pre-defined features. An essay that does not fit the rubric or has the right keywords may not get the highest score. Moreover, the rubric may not consider areas in which the student is strong. All machine learning systems are prone to error, and AES may make mistakes while scoring. For this reason, in high-stakes situations, we should give an AES system’s input to human reviewers for verification.
Using AES in Schools
AES is typically implemented as a component of a larger evaluation system. This can include multiple-choice exams and written exams. AES is also used to help teachers assign writing assignments. Teachers can use it to assign topics to students based on strengths and weaknesses. We can also use it to give more feedback to students on the quality of their essays.
Want to build AES for your EdTech platform?
We help EdTech companies build next-generation learning software. Contact our data science team if you want to create an automated essay scoring system that can serve your students. You can learn more about our educational data science and AI work here: https://www.playpowerlabs.com/data-science
Thank you for reading!