Iranian EFL lecturers report moderate AI-induced anxiety, with job displacement the strongest concern

A cross-sectional study of 450 English as a foreign language lecturers in Iranian higher education found moderate overall levels of artificial intelligence-induced anxiety, with job displacement emerging as the strongest concern. The researchers also identified gender differences in anxiety about technical proficiency and technological support, finding that female lecturers reported higher levels of anxiety in those areas.

The study examined lecturers employed at universities in Iran. Participants had been teaching EFL at tertiary level for at least two years and were acquainted with at least one AI system, including ChatGPT, Perplexity or Grok. The sample included 125 male and 199 female lecturers, with a mean age of 39.3 years and mean teaching experience of 9.5 years. Lecturers were selected through purposive and snowball sampling, and data were collected through Google Forms between March and June 2025.

Researchers adapted Liu and Liu’s 2025 21-item Scale of Artificial Intelligence Anxiety for the Iranian context. The scale measures five dimensions: technical proficiency, job displacement, technological support, student experience and research development. The adaptation involved forward-backward translation, reconciliation by a third bilingual expert, mapping of items to the original constructs and cultural adjustments. Cognitive interviews with 45 Iranian EFL lecturers assessed the cultural appropriateness and comprehensibility of the items, after which some items were modified. A panel of five experts produced a content validity index of 92%, and the instrument was piloted among 40 Iranian EFL lecturers to determine its general applicability and clarity.

Confirmatory factor analysis supported the proposed five-factor structure. The model fit indices were χ²/df = 1.74, CFI = 0.96 and RMSEA = 0.054. Standardized factor loadings ranged from 0.62 to 0.83 and were statistically significant at p < .001. Average variance extracted values ranged from 0.48 to 0.62, while composite reliability ranged from 0.79 to 0.88. Multi-group confirmatory factor analysis established configural, metric and scalar invariance between male and female lecturers, supporting the use of observed scores for gender comparisons.

On the five-point scale, the total AIA score averaged 3.45 (SD = 0.78), indicating moderate anxiety overall. Job displacement had the highest dimension mean, at 3.72 (SD = 0.85), followed by research development at 3.58 (SD = 0.82). Technical proficiency averaged 3.41 (SD = 0.79), student experience 3.38 (SD = 0.76) and technological support 3.16 (SD = 0.74). At the item level, concerns about AI replacing some teaching responsibilities and making research techniques obsolete received the highest mean scores.

Independent-samples t-tests found significantly higher anxiety among female lecturers for technical proficiency, 3.60 (SD = 0.81) compared with 3.22 (SD = 0.72) for male lecturers, with a Cohen’s d of 0.39. Anxiety about technological support was also higher among women, at 3.30 (SD = 0.76) versus 3.02 (SD = 0.69) among men, with d = 0.36. Women also reported higher anxiety for the overall AIA score. The analysis did not find statistically significant gender differences for job displacement, student experience or research development.

Moderated regression analyses, controlling for age and teaching experience, found that gender moderated the relationships between overall AIA and technical proficiency, with a significant interaction and an additional 4% of variance explained, and between overall AIA and technological support, with an additional 3% of variance explained. Simple-slope analysis indicated stronger associations for female lecturers in both areas. No significant gender interactions were found for job displacement, student experience or research development.

Structural equation modeling produced an acceptable fit for the higher-order AIA model, with χ²/df = 1.79, CFI = 0.95, TLI = 0.95, RMSEA = 0.056 and SRMR = 0.048. All five dimensions loaded strongly on the second-order construct, with coefficients from 0.72 to 0.88. Job displacement had the strongest loading, β = 0.88. Gender significantly moderated the associations with technical proficiency and technological support, but not with the other three dimensions. The authors interpret the selective pattern as evidence that AI-induced anxiety is multidimensional and that technical-related differences may be shaped by social and cultural conditions, including access to training and support.

The study has several limitations. Its cross-sectional design cannot establish how anxiety develops over time or support causal claims about AI, institutional support or professional outcomes. Gender was the only potential moderator examined, so factors such as digital literacy and teaching experience were not assessed as moderators. The findings are specific to Iranian EFL lecturers and the resource-constrained institutional context described in the study; they do not establish that the same pattern applies to other disciplines or countries. The authors recommend longitudinal research, comparisons across educational contexts and investigation of additional moderators, while the practical implications are framed as preliminary.

Subject of Research: AI-induced anxiety among Iranian English-language lecturers in higher education

Article Title: Psychological barriers and pedagogical implications of artificial intelligence-induced anxiety among English lecturers in higher education

Article References: Zahedkazemi, E., Khany, R., & Shoja, L. (2026). Psychological barriers and pedagogical implications of artificial intelligence-induced anxiety among English lecturers in higher education. Discover Education, 5(1), Article 939. https://doi.org/10.1007/s44217-026-02094-5

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02094-5

Keywords: artificial intelligence anxiety, higher education, Iranian EFL lecturers, job displacement, technical proficiency, technological support, gender differences, structural equation modeling, generative AI, Psychological, barriers, pedagogical