Investigating the performance of generative AI ChatGPT’s reading comprehension ability
The Korea English Education Society · 31.05.2023 · 22 (2) · 147-172
DOI: 10.18649/jkees.2023.22.2.147 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Korea English Education Society, online erschienen am 31.05.2023. Verfasst von Suh-Keong Kwon und 1 weiteren Autor:innen.
Abstract (Originalsprache)
This study examines the performance of the generative Artificial Intelligence, ChatGPT, in reading comprehension tasks of recent CSAT English tests. For this, several experiments were carried out in which ChatGPT responded to CSAT English reading items from the 2021 to 2023 versions, and the correct answer rates were determined. The study investigates how the correct answer rate might vary by item type and according to the different versions of ChatGPT (3.5 and 4). The findings reveal that ChatGPT-3.5 answered approximately 69% of the reading items correctly, while GPT-4 achieved almost perfect results. Additionally, the study explores whether similar trends are found in other foreign language tests, using TOEFL iBT reading items. The results indicate a similar trend in TOEFL iBT items. The implications for teachers, test developers, and language learners on how to use this generative AI for teaching and learning and for test-preparation purposes are discussed. The study highlights the potential of using AI in the language learning and teaching processes and its potential impact on test development.
Zitieren
APA
Kwon, S., & Lee, Y. T. (2023). Investigating the performance of generative AI ChatGPT’s reading comprehension ability. The Korea English Education Society, 22(2), 147-172. https://doi.org/10.18649/jkees.2023.22.2.147
Vancouver
Kwon S, Lee YT. Investigating the performance of generative AI ChatGPT’s reading comprehension ability. The Korea English Education Society. 2023;22(2):147-172. doi:10.18649/jkees.2023.22.2.147
BibTeX
@article{Kwon2023,
author = {Kwon, Suh-Keong and Lee, Young Tae},
title = {Investigating the performance of generative AI ChatGPT’s reading comprehension ability},
journal = {The Korea English Education Society},
year = {2023},
volume = {22},
number = {2},
pages = {147--172},
doi = {10.18649/jkees.2023.22.2.147}
}
EUVIMED