When AI threatens my future career: growth mindset and perceived employability in Chinese college students' career decision making anxiety

Synthesis: Chen surveyed 526 college students at one university in Zhejiang Province, China, to test a moderated mediation model linking AI employment threat perception to career decision-making anxiety, with perceived employability as the mediator and growth mindset as the moderator. AI employment threat perception predicted higher anxiety (b = 0.240) and lower perceived employability (b = −0.137), and the indirect path through perceived employability was significant (Effect = 0.042, 95% CI [0.013, 0.073]), leaving a still-significant direct effect (0.198) — partial mediation. Growth mindset weakened the threat-to-employability link: the interaction was significant (b = 0.122, p = 0.014) and the simple slope was significant only at low growth mindset (b = −0.229, p < 0.001) and not at high growth mindset (b = −0.013, p = 0.832). The design is a single-time-point, self-report, cross-sectional survey, so no causal or temporal direction is established and the authors say so explicitly.

Key Findings

• AI employment threat perception went with higher career decision-making anxiety. After controlling for gender, age, grade, major category, AI tool use frequency, and career planning course experience, the direct effect was b = 0.240, SE = 0.044, t = 5.385, p < 0.001 (H1).
• It also went with lower perceived employability. b = −0.137, SE = 0.043, t = −3.169, p = 0.002 (H2), and perceived employability in turn predicted lower anxiety, b = −0.303, SE = 0.043, t = −7.007, p < 0.001 (H3).
• Perceived employability partially mediated the link. Across 5,000 bootstrap resamples the indirect effect was Effect = 0.042, Boot SE = 0.015, 95% CI [0.013, 0.073]; with the mediator in the model the direct effect remained significant at 0.198, SE = 0.043, t = 4.608, p < 0.001, 95% CI [0.113, 0.284] (H4).
• Growth mindset buffered the threat-to-employability path. The AI employment threat perception × growth mindset interaction predicted perceived employability at b = 0.122, SE = 0.049, t = 2.473, p = 0.014 (H5).
• The simple slopes separate the two groups cleanly. At low growth mindset (−1 SD) the effect of AI employment threat perception on perceived employability was b = −0.229, SE = 0.060, t = −3.795, p < 0.001; at high growth mindset (+1 SD) it was b = −0.013, SE = 0.062, t = −0.213, p = 0.832.
• The indirect effect itself was conditional on growth mindset. The indirect effect through perceived employability was significant at low growth mindset (Effect = 0.069, Boot SE = 0.023, 95% CI [0.027, 0.119]) but not at high growth mindset (Effect = 0.004, Boot SE = 0.019, 95% CI [−0.033, 0.042]); the difference between the two conditional indirect effects was −0.065, Boot SE = 0.026, 95% CI [−0.119, −0.018], and the index of moderated mediation was −0.037, Boot SE = 0.017, 95% CI [−0.075, −0.008].
• All four constructs were empirically distinct. The four-factor measurement model fit well (χ²/df = 1.13, CFI = 0.988, TLI = 0.987, RMSEA = 0.016, SRMR = 0.026) and clearly outperformed the three-factor, two-factor, and single-factor models; Cronbach's alpha was above 0.90 for every construct and all HTMT values were below 0.85, with the highest at 0.672.

How the study was done

Participants were college students from a university in Zhejiang Province, China; 600 questionnaires were distributed and 526 valid responses were retained, an effective response rate of 87.67%. Among them 314 were female (59.7%) and 212 male (40.3%), with ages from 18 to 24 years (M = 20.32, SD = 1.31). All focal variables were measured on a seven-point Likert scale from 1 = strongly disagree to 7 = strongly agree, using items adapted from prior AI-anxiety, employability, career-anxiety, and growth-mindset research, with translation and back translation for scales originally developed in English. Item counts were 5 for AI employment threat perception, 5 for perceived employability, 5 for career decision making anxiety, and 4 for growth mindset. Because all measures were self-reported at a single time point, the authors report procedural remedies for common method bias and compare measurement models.

What this means for practice

• Target the appraisal, not just the reassurance. The authors argue that students' response to AI-driven change depends on how they read it; career education that gives occupation- and major-specific information about which tasks are exposed to automation and which human capabilities remain valuable is their first recommendation, aimed at replacing vague or exaggerated substitution fears with a more accurate picture.
• Strengthen perceived employability as the mediating resource. The indirect path ran through students' confidence in obtaining employment and adapting to labor-market change, so the authors pair counseling with concrete AI literacy training, digital skill development, interdisciplinary projects, internships, and practice applying AI tools to profession-specific tasks.
• Use growth-oriented feedback deliberately. Because the negative effect on perceived employability held only at low growth mindset, the authors recommend growth-oriented feedback, examples of successful skill transitions, structured reflection on learning progress, and individualized development plans — reinforcing the belief that career-relevant abilities can be developed.
• Screen for the high-risk combination. Students reporting simultaneously high AI employment threat, low perceived employability, and weaker growth mindset are, in the authors' account, a particularly vulnerable group worth identifying in career counseling, where accurate labor-market information, concrete skill-development opportunities, and psychological support may work better than general reassurance.
• Do not read the design as causal. Everything above is an association measured at one time point; the practical value is in what the mechanism suggests, not in a demonstrated causal chain.

Limitations

• Cross-sectional survey. The single-time-point, self-report design limits causal and temporal inference, and the paper is explicit that the temporal ordering among AI employment threat perception, perceived employability, and career decision making anxiety cannot be established; it notes that anxiety may itself shape how strongly students perceive AI-related employment risk, so reciprocal relationships remain possible.
• Self-report and same-source measurement. All focal variables came from the same respondents, so shared method variance cannot be fully eliminated, and AI employment threat perception captures subjective appraisal of technological risk rather than objective occupational exposure to AI. The authors suggest combining self-reports with external or behavioral indicators such as occupation- or major-specific AI exposure, AI-use behavior, skill-development activities, internship performance, or job-search outcomes.
• One institutional context in China. The sample came from students within a single institutional context, which limits generalizability; institutional, regional, cultural, and disciplinary differences may shape both perceived AI employment threat and career decision-making processes, so replication across universities, regions, countries, educational levels, and disciplines is needed.
• A partial model. The authors note the model covers only perceived employability and growth mindset, leaving out AI literacy, career adaptability, digital Self-Efficacy, social support, and future time perspective, and that it emphasizes threat appraisal although appraisal theory allows that change can also be read as manageable or as opportunity. Internal inconsistency in the reported correlations. The paper's two correlation tables disagree on several coefficients — for example the growth-mindset–career decision making anxiety correlation is −0.236 in the discriminant-validity table and −0.034 in the descriptive-statistics table — so the bivariate values should be read with care; the regression, mediation, and moderation estimates above are unaffected by this discrepancy.

Citation

Chen, H. (2026). When AI threatens my future career: growth mindset and perceived employability in Chinese college students' career decision making anxiety. Frontiers in Psychology, 17, 1933842.