Beyond Instant Answers: AI-Guided Inquiry in English Language Learning
Abstrak
The rapid spread of generative AI in English classrooms has made instant-answer use increasingly common, yet such use often shortens inquiry, weakens evidence checking, and reduces learner agency. This mixed-methods study examined whether an AI-guided inquiry model could move EFL learners beyond answer dependence toward question building, source verification, and reflective language use. Using an explanatory sequential design, 68 Indonesian university EFL students were assigned to an experimental class using a five-step AI-guided inquiry cycle and a comparison class using ordinary AI-assisted task completion for eight weeks. Quantitative data from an inquiry-oriented English learning rubric showed that the experimental group achieved a significantly higher mean gain (M = 18.74, SD = 8.10) than the comparison group (M = 8.29, SD = 6.09), t(66) = 6.01, p < .001, d = 1.46. The largest gains appeared in evidence integration and question formulation. Interview, log, and prompt-trace data further showed three changes: learners asked deeper questions, verified AI outputs before use, and retained stronger authorial control. The findings indicate that AI becomes pedagogically valuable when it is structured to guide inquiry rather than to supply instant answers.
Identifier
- DOI: https://doi.org/10.26858/eltww.v13i1.83972
- OpenAlex: https://openalex.org/W7164205530