AI evidence extraction

Lecturer's Perceptions and Strategies on ChatGPT Overreliance in ESL Academic Writing Among Undergraduates: A Case Study at a Malaysian Private University

Fairuz Umira Binti Azmi, Harwati Hashim · International Journal of Research and Innovation in Social Science · 2025

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

8/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.47772/ijriss.2025.91200050

Methodology & findings

Study design

Qualitative case study design with semi-structured interviews conducted with three ESL lecturers, analyzed using reflexive thematic analysis guided by the Theory of Planned Behaviour (TPB).

Sample

N = 3, 1 group

Primary method

Reflexive thematic analysis (qualitative analysis method, not statistical)

Main result

The study revealed that while ChatGPT offers linguistic scaffolding, lecturers perceived "a decline in authentic writing processes, diminished metacognitive engagement, and increasing occurrences of AI-generated inaccuracies and fabricated references." Additionally, "varied lecturer expectations and the lack of guidelines were found to encourage students' dependence on AI applications."

Reports effect sizes.

Research paradigm

Interpretivist/Qualitative

Author conclusions

The authors conclude that "these results are significant because they emphasise the institutional requirements for AI literacy education, unified governance and the restructuring of assessments to guarantee ethical and accountable AI application." They note that "this study contributes context-specific insights into sustainable AI integration aligned with SDG 4's call for quality education in the digital era."

Risk of bias

Selection bias: Small sample of only three lecturers from one institution; Potential interviewer bias in semi-structured interviews; Context-specific findings may not generalize beyond the Malaysian private university setting; Reflexive thematic analysis is subjective and dependent on researcher interpretation; Small sample size (n=3) - high risk of selection bias and limited representativeness; Single institution context - reduces generalizability; Potential interviewer bias in qualitative study with no inter-rater reliability measures mentioned; No mention of reflexivity protocols or peer debriefing to address researcher bias; Potential social desirability bias in lecturer responses about their teaching strategies; Small sample size (n=3 lecturers); Single institution case study (limited geographic/institutional diversity); Potential selection bias in lecturer recruitment; Interviewer bias in qualitative data collection

Open questions raised

The study identifies the need for institutional AI literacy education, unified governance frameworks for AI use, and restructured assessment methods. Future research directions include investigating how to balance AI scaffolding with authentic skill development and examining institutional-level policy responses.
The study implies gaps regarding: (1) need for institutional AI literacy education programs; (2) lack of unified governance guidelines for AI use in academic writing; (3) need for restructured assessment methods that verify authenticity; (4) limited research on lecturer perceptions of AI overreliance in ESL contexts at Malaysian institutions.
The authors identify the need for institutional AI literacy education, unified governance frameworks, and restructured assessment strategies to manage ChatGPT overreliance in ESL academic writing contexts.

Data: not_statedCode: not_statedExtracted from: pdfAgreement 80%

Explore related topics

AI Manuscript Writing Academic Writing Case Study