Assessing artificial intelligence and professors’ calibration in English as a foreign language writing courses at a Costa Rican public university

Abstract

This article paper explores the evaluation of artificial intelligence (AI) in English as a Foreign Language (EFL) writing courses and the importance of calibration in writing evaluations. The role of calibration has received little attention in language contexts, while the role of artificial intelligence has gained increased attention in the last couple of years. This investigation, conducted from August 2022 to March 2023, involved eight TESOL students enrolled in an English as a Foreign Language (EFL) major at a Costa Rican public university, ten TESOL university professors, and one AI piece of software. It used a quantitative, quasi-experimental design, and a language elicitation data collection process. Data was collected by means of a rubric-based writing assessment. Quantitative data were analyzed using descriptive statistics. Data analyses indicate that: 1) human-created paragraphs (X̄ = 7,56) and AI writing (X̄ = 7,61) yield similar results when evaluated; 2) some criteria may favor human creativity or computer, rule-oriented writing; and 3) professors’ ratings reveal inconsistencies when grading human writing in particular. These findings demonstrate that AI matches, at least to a basic level, human writing skills. Furthermore, data show that students may be falling behind in aspects such as grammar, vocabulary, and mechanics. Finally, the analysis indicates that professors’ grading lacks consistency, and a calibration model should be incorporated as part of regular training workshops.