Working with AI: the impact of organizational intelligent service strategy on employees’ perception of career achievement

The study examines how AI adoption in service organizations influences front-line employees’ perception of career achievement, an intrinsic motivation reflecting how employees evaluate their work experiences and attainments. Prior findings conflict: AI can threaten jobs and reduce perceived achievement, or improve efficiency and service quality, enhancing achievement. The authors propose that the contradiction can be explained by organizations’ intelligent service strategies—intelligence substitution (AI replaces frontline roles) versus intelligence collaboration (AI augments employees). They pose two research questions: (RQ1) What mechanism explains the differential impacts of these strategies on employees’ perception of career achievement? (RQ2) Do these impacts depend on the organizational innovation climate? Grounded in self-perception theory, the study argues that changes in work patterns and interactivity (human-human vs human-machine) shape employees’ intrinsic perceptions, and that organizational innovation climate can alter how employees interpret and react to AI-driven changes.

Perception of career achievement is influenced by internal factors (e.g., gender, age, self-efficacy, job insecurity, workload) and external factors (e.g., organizational climate and work methods). Existing research has largely addressed traditional service settings and focused on job insecurity and performance impacts of AI, overlooking how AI-induced changes in work patterns affect intrinsic motivation. Self-perception theory posits that individuals infer internal states from their behaviors and the contexts in which they occur; thus, shifts from human-human to human-machine interaction can trigger new self-attributions affecting perceived career achievement. Intelligent service strategies are conceptualized as: (1) intelligence substitution, where AI dominates service delivery and employees are displaced to back-office roles, and (2) intelligence collaboration, where AI assists employees, preserving and enhancing employee-customer interactions. Interactivity is defined as employees’ perception of interaction forms in the service environment and is split into human-human interactivity (employee-customer, face-to-face) and human-machine interactivity (employee/customer with AI). Hypotheses: H1—substitution (vs collaboration) lowers perceived career achievement. H2a—substitution lowers human-human interactivity. H2b—substitution increases human-machine interactivity. H3a—human-human interactivity positively affects perceived career achievement. H3b—human-machine interactivity negatively affects perceived career achievement. The organizational innovation climate (OIC), reflecting support for new technologies and methods, is proposed to moderate these links by weakening the positive effect of human-human interactivity and mitigating the negative effect of human-machine interactivity on perceived career achievement (H4a, H4b).

Three studies with frontline employees tested the hypotheses across experience- and credence-based services. - Study 1: Online experiment with 223 frontline employees (58.7% female; Mage=32.38) recruited via Credamo. Single-factor between-subjects design manipulating intelligent service strategy (substitution vs collaboration) in a fictitious retail company. Participants imagined working as supermarket clerks, read a strategy description, wrote reflections (up to 5 minutes), then completed measures of human-human interactivity (5 items), human-machine interactivity (5 items), and perception of career achievement (6 items), all on 7-point Likert scales. Manipulation checks assessed perceived substitution vs collaboration. Controls: gender, education, AI-use experience. CFA supported reliability/validity (e.g., CR>0.8, AVE>0.5; fit: χ²=136.22, χ²/df=1.35, GFI=0.93, CFI=0.99, RMSEA=0.04). Analyses: ANOVA/MANOVA, regressions, and mediation (PROCESS Model 4, 5000 bootstraps). - Study 2: Online experiment with 232 frontline employees (63.4% female; Mage=30.32) via Credamo. 2×2 between-subjects design: intelligent service strategy (substitution vs collaboration) × organizational innovation climate (high vs low) in a fictitious insurance firm. Procedure analogous to Study 1, with added manipulation of OIC (high vs low) and subsequent OIC measurement. CFA supported measurement (χ²=174.49, χ²/df=1.73, GFI=0.92, CFI=0.97, RMSEA=0.06). Analyses: ANOVA/MANOVA, regressions, mediation (PROCESS Model 4), and moderated mediation (PROCESS Model 14, 5000 bootstraps). - Study 3: Field-based experiment with 281 frontline employees (63.7% female; Mage=31.69) from a provincial branch of the Agricultural Bank of China. 2×2 between-subjects design (service strategy × OIC) with realistic banking descriptions. Measures and manipulations as in Study 2, plus demographics and manipulation checks. Analytic approach used PLS-SEM for measurement and structural paths (moderation modeled via product indicators), with control variables (gender, education, AI-use experience). Validity supported (most loadings >0.7; CR>0.8; AVE>0.5; HTMT<0.85). Structural model evaluated path coefficients, R², Q², and indirect effects; multigroup analysis compared low vs high OIC groups.

Across three studies (total N=736; valid Ns=223, 232, 281), results consistently supported all hypotheses. - Study 1 (retail; N=223): Perception of career achievement (PCA) was lower under substitution (M=4.46, SD=1.44) than collaboration (M=5.59, SD=0.74), F(1,221)=52.76, p<0.001, partial η²=0.19 (H1). Substitution reduced human-human interactivity and increased human-machine interactivity (e.g., HH: F(1,221)=270.24, p<0.001; HM: F(1,221)=50.47, p<0.001) (H2a,b). Regression: HH→PCA b=0.35, SE=0.05, p<0.001 (H3a); HM→PCA b=-0.26, SE=0.11, p<0.05 (H3b). Mediation (PROCESS Model 4): direct ISS→PCA not significant (b=0.19, SE=0.21, 95% CI [-0.23,0.61]); indirect via HH significant (b=0.76, SE=0.18, 95% CI [0.41,1.13]) and via HM significant (b=0.17, SE=0.07, 95% CI [0.06,0.33]). - Study 2 (insurance; N=232): PCA lower under substitution (M=4.84, SD=1.38) vs collaboration (M=5.32, SD=1.16), F(1,230)=8.31, p<0.01 (H1). Substitution lowered HH and raised HM interactivity (HH: F(1,230)=100.93, p<0.001; HM: F(1,230)=49.19, p<0.001) (H2a,b). HH→PCA positive (b=0.20, SE=0.06, p<0.01) and HM→PCA negative (b=-0.21, SE=0.09, p<0.05) (H3a,b). Mediation: direct ISS→PCA ns (b=-0.07, SE=0.19, 95% CI [-0.45,0.30]); indirect via HH (b=0.35, SE=0.15, 95% CI [0.07,0.66]) and via HM (b=0.20, SE=0.10, 95% CI [0.02,0.42]) significant. OIC moderated effects (PROCESS Model 14): HH×OIC b=-0.22, SE=0.11, 95% CI [-0.44,-0.01] (weakens HH→PCA); HM×OIC b=0.40, SE=0.16, 95% CI [0.09,0.72] (reduces HM’s negative impact) (H4a,b). ISS×OIC ANOVA interaction: F(1,228)=8.91, p<0.01; under substitution, PCA higher with high OIC (M=5.60) vs low OIC (M=4.09); under collaboration, PCA also higher with high OIC (M=5.65) vs low OIC (M=5.04). - Study 3 (banking; N=281): PLS-SEM showed ISS increased HH (b=0.78, SD=0.11, p<0.001) and decreased HM (b=-0.69, SD=0.10, p<0.001) (H2a,b). HH positively (b=0.60, SD=0.11, p<0.001) and HM negatively (b=-0.30, SD=0.10, p<0.01) affected PCA (H3a,b). Total ISS→PCA effect significant (b=0.73, SD=0.13, p<0.001); direct effect ns (b=0.06, SD=0.10, p=0.56). Indirect via HH (b=0.47, SD=0.10, p<0.001) and via HM (b=-0.20, SD=0.07, p<0.01) significant (H1 via mediation). OIC moderated HH→PCA (b=-0.29, SD=0.14, p<0.05) and HM→PCA (b=0.26, SD=0.11, p<0.05) (H4a,b). Model R² for PCA=0.41; Q²=0.10. Multigroup analysis: effects stronger under low OIC (HH→PCA b=0.58 vs 0.32; HM→PCA b=-0.29 vs -0.05; mediated paths significantly larger in low OIC group).

Findings demonstrate that organizational intelligent service strategy shapes front-line employees’ perception of career achievement through changes in perceived interactivity patterns, aligning with self-perception theory. Intelligence substitution decreases human-human and increases human-machine interactivity, which respectively increases and decreases perceived career achievement, yielding a net decline relative to collaboration. Organizational innovation climate functions as a boundary condition: it weakens the positive influence of human-human interactivity and cushions the negative impact of human-machine interactivity on perceived career achievement. These results clarify why prior studies reported conflicting outcomes of AI adoption on employees—differences in strategy (substitution vs collaboration) and organizational climate yield divergent psychological consequences. The work contributes a multilevel framework connecting organizational strategy, work pattern perceptions, and individual intrinsic motivation in AI-enabled service contexts.

The paper advances understanding of how AI-enabled service strategies affect employees’ intrinsic motivation. Across three studies, intelligence substitution (vs collaboration) lowered front-line employees’ perception of career achievement primarily via reduced human-human and increased human-machine interactivity. Human-human interactivity boosted, while human-machine interactivity dampened, perceived achievement. Organizational innovation climate moderated these pathways, attenuating both the positive HH→PCA link and the negative HM→PCA link. Contributions include: introducing intelligent service strategy as a determinant of intrinsic motivation; identifying interactivity perceptions as mediators; and establishing organizational innovation climate as a critical boundary condition. Practically, organizations aiming to implement AI should prefer collaboration strategies and cultivate an innovation climate to buffer adverse effects on employees. Future research should refine interactivity taxonomies (e.g., triadic customer–AI–employee interactions), incorporate customer feedback impacts on employee perceptions, and examine individual differences such as psychological job demand in moderating AI’s effects.

The study classifies interactivity broadly as human-human and human-machine; richer typologies (e.g., dynamic triadic consumer–AI–employee interactions) may capture additional variance. Employee perceptions may be influenced by external feedback, particularly customer reactions to AI services, which were not explicitly modeled. Individual differences (e.g., psychological job demand, variability in AI-related skills) may shape how employees interpret and adapt to AI-driven changes. Experimental manipulations relied on scenarios and field descriptions that, while validated, may not capture all real-world complexities across industries and cultures.