ChatGPT Employment Anxiety and Career Confidence: Discipline-Contingent Evidence From Malaysian Undergraduates

Authors

• Fahd A. Raza, Asia Pacific University of Technology & InnovationFollow
• Abtar D. Singh, Asia Pacific University of Technology & InnovationFollow
• Rabia Anwar, Asia Pacific University of Technology & InnovationFollow
• Jonathan JS Kovilpillai, Asia Pacific University of Technology & InnovationFollow
• Analisa Hamdan, Asia Pacific University of Technology & InnovationFollow
• Fumiko Konno, Asia Pacific University of Technology & InnovationFollow
• Husna HR Azami, Asia Pacific University of Technology & InnovationFollow
• Murali Raman, Asia Pacific University of Technology & InnovationFollow

Digital Object Identifier

10.18870/hlrc.v16i1.1878

ORCID

Fahd Ali Raza, https://orcid.org/0000-0003-4577-6083;

Prof. Abtar Darshan Singh, https://orcid.org/0000-0003-4525-5966;

Rabia Anwar, https://orcid.org/0009-0004-3672-4825;

Jonathan Jeevan Strinivas Kovilpillai, https://orcid.org/0000-0001-7517-2809;

Analisa Hamdan, https://orcid.org/0000-0003-4225-129X;

Dr. Fumiko Konno, https://orcid.org/0009-0009-4510-9142;

Dr. Husna Hafiza R. Azami, https://orcid.org/0000-0002-2845-2004;

Prof. Murali Raman, https://orcid.org/0000-0003-2392-628X  

Abstract

Background: The rapid diffusion of generative artificial intelligence (AI) has raised concerns about graduate employability, yet evidence on how ChatGPT employment anxiety relates to career confidence remains limited, particularly in non-Western contexts.

Objectives: We examined whether ChatGPT employment anxiety is associated with career confidence among Malaysian undergraduates and whether academic discipline or usage intensity moderates that association.

Methods: Using the Malaysian subset of the 2024–2025 Global ChatGPT Student Survey (Aristovnik et al., 2025; N = 315 ChatGPT users), we constructed a four-item employment anxiety index (α = .80) and estimated hierarchical ordinary least squares (OLS) regressions predicting a single-item measure of career confidence, controlling for age, gender, institution type, study level, discipline, and usage intensity. Hierarchical regression entered predictors in successive blocks so that the variance explained by employment anxiety, beyond the control variables, could be assessed. We used regression diagnostics and a multiple-imputation sensitivity analysis to check model assumptions and confirm that results did not depend on how missing data were handled.

Results: Contrary to H1, employment anxiety showed no overall association with career confidence (β = .029, p = .562). However, an exploratory, discipline-specific pattern emerged (H2 partially supported): The anxiety × information technology (IT) and engineering interaction was significant (β = −.311, p = .005). Within this group, higher anxiety was associated with lower confidence at the conventional threshold (β = −.164, p = .050), whereas no such relationship appeared among business or arts, humanities, and natural sciences students. Usage intensity did not moderate the relationship (H3 not supported). Findings were robust to multiple imputation, meaning the study’s final conclusions did not change.

Conclusions: These exploratory findings suggest that the link between AI employment anxiety and career confidence may be discipline contingent rather than uniform—a pattern that does not completely align with Western automation-anxiety models. We framed the buffering role of Malaysia’s collectivist context as an interpretive hypothesis rather than a tested claim because we did not measure cultural orientation. The discipline-specific association among IT and engineering students, who face the most direct occupational exposure, warrants replication. Also, the cross-sectional design and single-item outcome measure limit causal and practical inference.

Recommended Citation

Raza, F. A., Singh, A. D., Anwar, R., Kovilpillai, J. J., Hamdan, A., Konno, F., Azami, H. H., & Raman, M. (2026). ChatGPT Employment Anxiety and Career Confidence: Discipline-Contingent Evidence From Malaysian Undergraduates. Higher Learning Research Communications, 16 (2).
DOI:10.18870/hlrc.v16i1.1878