Speaker
Description
The widespread adoption of Large Language Models (LLMs) is reshaping the teaching and learning of literary reading and interpretation, while simultaneously highlighting the need to develop learners’ AI evaluation and AI critique competencies. This study empirically examines the capabilities of LLMs in identifying and analyzing similes through a case study of the poetry of Bình Nguyên Trang. The corpus was compiled from two poetry collections, and the models’ outputs were evaluated against a researcher-developed reference key. The findings indicate that LLMs perform well in identifying similes with explicit formal markers but become less reliable when interpreting their contextual meanings. They also exhibit tendencies toward formulaic interpretation and unsupported inference. Based on these findings, the study argues that LLM-generated responses should be treated as analytical hypotheses requiring critical verification rather than authoritative interpretations, thereby contributing to the development of learners’ AI critique competencies in literature education.
Keywords: AI literacy, AI critique, Large Language Models (LLMs), simile
