AI Awareness and Employee Innovation: A Dual-Pathway Moderated Mediation Model Within Organizational Systems

Abstract

Amid AI-driven organizational transformation, employees’ subjective evaluations of AI technologies—referred to as AI awareness—have become a critical psychological factor influencing innovation behavior. This study aims to uncover how AI awareness affects employee innovation performance through behavioral pathways and examines the moderating role of individual achievement motivation in this process. Grounded in Cognitive Appraisal Theory and the Dual Pathway Model, we construct a dual-path mediation model, in which proactive and withdrawal behaviors act as mediators, and achievement motivation serves as a boundary condition. Based on a two-wave survey of 413 knowledge workers in China’s high-tech sector, the proposed model was empirically tested using moderated mediation analysis (PROCESS macro). The results reveal that AI awareness has a significant dual-path effect on innovation behavior: on the one hand, it promotes innovation by stimulating proactive behavior; on the other hand, it may suppress innovation by inducing withdrawal behavior. Moreover, achievement motivation plays a crucial moderating role in this mechanism by strengthening the positive mediating effect of proactive behavior and weakening the negative mediating effect of withdrawal behavior. This study conceptualizes AI awareness as a psychological input encompassing both perceived opportunity and perceived threat, clarifies the behavioral response mechanisms of employees facing AI, and highlights the boundary-regulating role of individual motivation in organizational adaptability. Practically, the study suggests designing differentiated HR interventions based on employees’ cognitive appraisals and motivational profiles to enhance human–AI collaboration, foster innovation resilience, and improve organizational adaptability amid digital transformation.

1. Introduction

Amid the rapid advancements of Artificial Intelligence (AI) technologies, the Fourth Industrial Revolution is fundamentally transforming business operations and organizational structures at an unprecedented pace. This transformation is not only reflected at the technological level but also profoundly affects the coordination structure and operational logic among various internal elements of the organization. An organization can be viewed as a management system composed of multiple interrelated functional units, within which employee behavior represents one of the most sensitive and critical components. In the context of AI-driven transformation, changes in employee behavioral mechanisms directly influence the responsiveness and adaptability of the organizational system [1].

Compared with traditional technologies, AI features human-like intelligence and demonstrates high efficiency and adaptability in information processing, learning feedback, and automated decision-making. It has been widely adopted in industries such as manufacturing, healthcare, finance, and services [2]. In addition to driving continuous innovation in products and services, AI fundamentally alters career structures and job roles, particularly in high-tech enterprises, where organizational operations are significantly dependent on technological innovation [3]. Consequently, employees increasingly face occupational uncertainty and mounting pressure to reskill.

Although the implementation of AI has brought significant efficiency gains to organizations, it has also raised employee concerns about job displacement, task restructuring, and blurred role boundaries [4]. AI can perform highly structured and repetitive tasks and serves as a powerful tool to improve quality and efficiency. However, its limitations in emotional interaction and empathy complicate human–machine collaboration, placing additional emotional and adaptive demands on employees. Under these circumstances, awareness of employees regarding AI, that is, their subjective perception of AI, has emerged as a key psychological factor influencing their mental states and behavioral responses [4]. AI awareness reflects employees’ cognitive appraisals of how AI technologies, algorithmic systems, and data-driven management may affect their career development, skill relevance, and perceived job value [5]. Prior research has suggested that AI awareness may stimulate proactive behaviors but may also induce anxiety, emotional exhaustion, or withdrawal, thus shaping employees’ adaptive capacities in the workplace [6,7].

While AI awareness has garnered increasing academic attention, most studies have focused on its effects on stress [8,9], emotional responses [10], and work attitude [5,11]. Few studies have systematically examined how AI awareness affects employee innovation behavior, which is a critical driver of performance. Innovation behavior refers to the generation, promotion, and implementation of novel ideas in the workplace and plays a crucial role in achieving competitive advantage and sustaining organizational renewal, particularly in the context of digital transformation [12,13]. In contexts characterized by high uncertainty and technological transformation, the motivational mechanisms underlying innovation behavior become increasingly complex. It remains to be thoroughly explored whether employees, when confronted with AI-related stimuli, exhibit innovation suppression or activation depending on their cognitive appraisals. Therefore, clarifying how AI awareness influences employee innovation behavior through specific psychological mechanisms can not only fill the existing research gap at the behavioral level but also help organizations formulate more motivating human resource management strategies during AI-driven transformation.

This study focuses on knowledge workers due to their central role in driving innovation and their heightened sensitivity to AI-related changes [14]. These employees rely heavily on professional expertise, analytical judgment, and cognitive skills and are often among the first to experience the effects of task redefinition, skill obsolescence, and role ambiguity triggered by AI [15]. Moreover, behavioral responses have a direct impact on organizational innovation and strategic agility [16,17]. Therefore, examining the ways in which AI awareness influences innovation behavior among knowledge-based employees has both theoretical and practical implications.

To explore this relationship, this study draws on the Cognitive Appraisal Theory and the Dual Pathway Model. The former posits that individuals evaluate external stimuli based on available resources and personal experiences: when perceived as a “challenge stressor,” a stimulus may elicit positive coping responses; when viewed as a “hindrance stressor,” it may lead to avoidance or withdrawal [18]. The latter suggests that individuals under pressure typically adopt one of two competing behavioral pathways, proactive behavior or withdrawal behavior, which consumes limited personal resources [19]. Based on this, the present study posits that AI perception, as an external situational variable, activates individuals’ cognitive appraisals, thereby triggering either proactive or avoidance behavioral pathways, which ultimately influence their innovative behavior.

Moreover, when individuals encounter emerging technologies such as AI, their cognitive appraisals of external stimuli and corresponding behavioral responses are not solely dictated by the objective attributes of the technologies themselves but are profoundly shaped by internal psychological resources [18]. Among various personal traits, achievement motivation—defined as the intrinsic drive to pursue success and avoid failure—emerges as a particularly critical factor [20]. Compared to self-efficacy, which emphasizes individuals’ belief in their capabilities, or autonomy orientation, which highlights behavioral independence, achievement motivation more directly influences how individuals appraise challenges and choose behavioral strategies under conditions of uncertainty [21]. Individuals with high achievement motivation tend to perceive AI-driven changes as opportunities for growth and respond proactively, whereas those with low achievement motivation are more likely to view AI as a threat and exhibit defensive or avoidant behaviors.

Therefore, this study introduces achievement motivation as a moderating variable to examine its role in moderating the relationship between AI perception and innovative behavior through different behavioral pathways.

Based on the above, this study addresses the following core research questions: How do employees’ subjective perceptions of AI affect their innovation behavior in organizations undergoing digital transformation? Do proactive and withdrawal behaviors constitute dual pathways through which AI awareness influences innovation? How does achievement motivation moderate this relationship? To address these questions, this study developed and empirically tested a moderated mediation model in which AI awareness affects employees’ innovation behavior through both proactive and withdrawal behaviors, with achievement motivation as the key moderating variable.

The theoretical contribution of this study lies in integrating the cognitive appraisal theory of stress with the dual-pathway model to uncover the psychological mechanisms and behavioral strategies of employees in AI-related contexts, thereby enriching the research perspective in the field of AI and organizational behavior. In practice, the study offers actionable managerial insights for organizations to develop differentiated human resource strategies and targeted intervention and incentive systems during the process of AI-driven transformation.

2. Theory and Hypotheses

2.1. Challenge and Hindrance Stressors

As research on organizational behavior has continued to evolve, work-related stress has become a key area of interest for scholars [22]. Due to the widespread presence of stressors in the workplace, researchers have increasingly examined their effects on employee behavior and psychological states, aiming to develop more effective strategies for managing workplace stress.

Broadly defined, stressors refer to events or characteristics within the work environment that can elicit psychological tension or strain in employees, often serving as precursors to adverse psychological or physiological states [23]. Numerous scholars have suggested that stressors are consistently linked to negative outcomes, such as perceived stress, anxiety, health issues, turnover intentions, fatigue, emotional exhaustion, and counterproductive work behaviors, while being negatively associated with positive outcomes, such as job performance [10,24,25].

However, other researchers have contended that stressors are not inherently harmful. Some studies have found that certain stressors can produce positive outcomes, such as increased work engagement, well-being, and performance levels [26]. Selye [27] was the first to propose that stressors possess “dual attributes” and emphasized that both low and high levels of stress could have significant impacts on individuals. This perspective aligns with Cognitive Appraisal Theory, proposed by Lazarus [18], which posits that individuals’ emotional and behavioral responses to external stimuli are shaped by their subjective cognitive interpretations—particularly whether they perceive the stressor as a challenge or a hindrance.

On this basis, Cavanaugh et al. [28] introduced the “Challenge–Hindrance Stressor” model, building on Lazarus’s theory. In this model, challenge stressors refer to pressures conducive to personal growth and career development, including high workload, time pressure, and job responsibility. Although such stressors may temporarily deplete resources, successful coping can lead to psychological satisfaction, enhanced capabilities, and career benefits. In contrast, hindrance stressors are those perceived as insurmountable and obstructive to goal achievement or personal growth—such as role conflict, interpersonal tensions, and career stagnation. These stressors consume significant personal resources and offer minimal returns, often resulting in adverse outcomes [29]. This model expands the theoretical understanding of the dual effects of stress and provides a foundational framework for explaining individuals’ diverse behaviors in complex organizational contexts.

This study conceptualizes AI awareness as an external stressor within organizational settings and applies the challenge–hindrance stressor framework for analysis. Employees’ perceptions of AI vary significantly: some regard it as a growth-promoting challenge and respond with active adaptation and innovation-seeking behaviors, while others perceive it as a threat, leading to stress or avoidance. These differing cognitive appraisals result in two relatively distinct behavioral pathways: a proactive pathway characterized by engagement and opportunity-seeking, and a defensive pathway marked by withdrawal and self-protection. Thus, this theoretical framework not only helps to elucidate the cognitive processing and behavioral decision-making mechanisms employees employ in response to technological change but also provides a robust basis for understanding the diversity of innovation behaviors in the context of AI.

2.2. The Dual Behavioral Pathways of AI Awareness: Mechanisms of Influence on Employees

AI awareness refers to employees’ subjective evaluations of how AI and related technologies affect their career development, skill adaptability, and task content [4,5]. This perception reflects employees’ anticipatory judgments about the opportunities and threats posed by AI and reveals their psychological readiness and behavioral tendencies in response to technological change [30].

Employees’ work behaviors can be broadly categorized into two types: proactive and withdrawal. Proactive behavior refers to employees’ self-initiated actions undertaken without explicit external commands aimed at improving work processes or enhancing performance. This includes behaviors such as self-directed learning, offering constructive suggestions, and assuming extra responsibility [31]. Contrastingly, withdrawal behavior encompasses avoidant behaviors that employees adopt to cope with pressure, including reduced work effort, task avoidance, and disengagement from organizational activities [32]. At its core, it is a form of psychological defense.

Against the backdrop of rapid AI-driven transformation of organizational management and operational workflows, employees’ subjective evaluations of AI significantly shape their behavioral response paths. Employees who perceive AI as a means of skill enhancement, learning, and career advancement are more likely to appraise it as a challenge stressor [33]. This evaluation activates work motivation and encourages employees to invest in acquiring human–AI collaboration skills, optimizing workflows, and engaging in organizational transformation, thereby demonstrating higher levels of proactive behavior [34]. In this context, employees view the stress induced by AI technology as a valuable opportunity for achieving personal goals and realizing self-worth, rather than a threatening disruption. Such challenge appraisals intrinsically motivate employees to focus on tasks and assume additional responsibilities [35]. Moreover, the empowerment effects of AI, such as increased efficiency, freed-up time, and cognitive resources, enable employees to plan their work proactively, enhance their capabilities, and foster innovation. Similarly, Liang et al. [11] emphasize that employees who perceive AI as providing greater job flexibility and resource support are more inclined to adopt self-management strategies and continuous improvement to enhance role fit and irreplaceability.

Conversely, in the context of AI’s ongoing penetration of AI into enterprise work scenarios, if employees perceive that AI increases task complexity, raises skill demands, or jeopardizes job stability, they may evaluate it as a threat that can elicit negative behavioral responses [24]. Specifically, when employees face the uncertainty and pressure associated with AI awareness and believe that they lack the capacity or resources to effectively respond to technological change, they may adopt withdrawal behavior as a cognitive avoidance and psychological defense strategy [36]. Such behaviors include reduced work involvement, avoidance of technical tasks, passive compliance, and even emotional absenteeism [37], which essentially attempt to minimize presence and visibility to avoid potential failure or loss [38].

As AI deployment accelerates across industries, employees’ perceived uncertainty about their career trajectories also intensifies. AI awareness often triggers a sense of potential replacement concerns that their positions may be marginalized or eliminated by technological advancement, leading to anxiety, job insecurity, and other negative psychological reactions. Adapting to new AI-driven tasks, systems, and workflows imposes additional cognitive demands and learning burdens, further reinforcing employees’ threat-based appraisals of AI [39]. Prior research has identified job insecurity as a major antecedent of withdrawal behavior, with AI serving as a critical trigger of such stress [40]. When employees view AI awareness as a threat to their professional future, their appraisal aligns with hindrance stress, and they are more likely to exhibit defensive withdrawal behaviors. These behaviors diminish their work motivation and performance and erode their willingness to engage in organizational innovation.

In summary, AI awareness as a complex psychological variable possesses the dual attributes of both challenge and hindrance. Its ultimate effect on employee work behavior depends on how employees subjectively evaluate AI and which psychological coping mechanisms are activated in response. Based on this, the following research hypothesis is proposed:

Hypothesis 1a.

AI awareness has a significantly positive effect on employees’ proactive behavior.

Hypothesis 1b.

AI awareness has a significantly positive effect on employees’ withdrawal behavior.

2.3. The Mediating Role of Proactive and Withdrawal Behaviors

Proactive behavior refers to employees’ self-initiated, improvement-oriented actions based on acute insight into the work environment and taken without external compulsion. This behavior is characterized by foresight, adaptability, and constructiveness [31,41]. It includes activities such as voluntarily learning new skills, optimizing workflows, proposing suggestions for improvement, and actively participating in organizational change. Especially in the context of AI awareness, employees who perceive AI as a resource for personal growth are more likely to engage in human–AI collaborative learning, AI system operations, and digital workflow optimization, thereby continuously stimulating their innovative potential and performance [42].

From a cognitive perspective, proactive behavior enhances employees’ sense of task control and self-efficacy, making them more confident and motivated when faced with work challenges [19]. Furthermore, it fosters resource integration and stimulates diverse thinking through knowledge acquisition and cross-functional collaboration [43], thus providing a solid knowledge base for generating new ideas. At the organizational level, proactive behavior contributes to improving work environments and process efficiency, cultivating a supportive climate for exploration and innovation [41]. This, in turn, reduces the cost and risk of innovation, increasing employees’ willingness to experiment and take the initiative. Therefore, proactive behavior is a positive response to challenge stressors, as well as a key internal mechanism for activating innovative behavior.

By contrast, withdrawal behavior reflects a negative coping strategy that arises when employees feel incapable of managing stress or lack adequate resources. At its core, it is a form of cognitive avoidance and psychological defense [44]. These include behaviors such as avoiding complex tasks, delaying feedback responses, complying passively with directives, lowering task commitment, and engaging in emotional absenteeism [45]. In the context of AI’s increasing organizational penetration of AI, employees who perceive AI as a threat to their job security or competencies and fail to find effective coping mechanisms are more likely to evaluate AI as a hindrance stressor. Consequently, they may adopt withdrawal behaviors to “lower their visibility” and avoid the potential negative outcomes associated with uncertainty [36].

Moreover, withdrawal behavior dampens employees’ initiative and exploratory drive. Additionally, it leads to cognitive rigidity, emotional disengagement, and avoidance of collaboration during innovation tasks, distancing them from creative work environments. Over time, such behavior can result in emotional exhaustion, cognitive fatigue, and professional burnout [46] while also eroding organizational identification, team cohesion, and a sense of responsibility. This ultimately undermines the organization’s innovation–supportive climate and disrupts the configuration of social and knowledge-sharing networks [47].

In summary, proactive and withdrawal behaviors represent two typical behavioral pathways activated by employees in response to AI-related stress. They correspond to a constructive mechanism for creative engagement and a defensive mechanism for self-protection, respectively, and constitute critical mediating channels through which AI affects innovative employee behavior. Accordingly, the following research hypothesis is proposed:

Hypothesis 2a.

Proactive behavior plays a positive mediating role in the relationship between AI awareness and employee innovative behavior.

Hypothesis 2b.

Withdrawal behavior plays a negative mediating role in the relationship between AI awareness and employee innovative behavior.

2.4. The Moderating Role of Achievement Motivation

Achievement motivation is widely regarded as a key internal driver that compels individuals to set goals, exert effort, and pursue excellence in performance [48]. Atkinson divided achievement motivation into two dimensions: the motive to approach success and the motive to avoid failure. The former triggers proactive behaviors, whereas the latter is more likely to lead to avoidance and defense behaviors. Therefore, achievement motivation shapes cognitive processing of external information in individuals and influences their behavioral response pathways under stress.

Among individuals with high achievement motivation, external changes such as the introduction of AI are more likely to be perceived as opportunities for self-enhancement and capability development [49]. These individuals are characterized by a strong goal orientation and a heightened sense of control, making them more willing to embrace challenges and engage proactively to gain a sense of achievement. When confronted with work adjustments or skill transformation demands brought about by AI, highly achievement-motivated employees are more inclined to redefine tasks and upgrade competencies to strengthen their strategic positioning within AI-driven organizational structures. For example, they may voluntarily apply for technical training, participate in system development, or propose workflow optimization initiatives. This goal-oriented behavioral drive forms the psychological basis of proactive behaviors [50,51].

In contrast, employees with low achievement motivation generally lack a strong desire for success and are more driven by a “failure avoidance” orientation. They typically avoid uncertainty and experience greater anxiety about failure [52]. When AI implementation introduces skill obsolescence, increased task complexity, or ambiguity in career progression, these individuals are more likely to experience helplessness and cognitive threat, prompting them to adopt a “deactivation” strategy to minimize the risk of failure. For instance, they may avoid operating new systems, resist taking on technical tasks, reduce work engagement, or even display withdrawal behaviors, such as emotional absenteeism or passive compliance [53]. These behaviors constitute a “resource-conserving” defense mechanism aimed at avoiding negative emotions or organizational penalties by lowering visibility and participation.

According to the Cognitive Appraisal Theory of Stress [18], individuals’ subjective evaluations and coping strategies when faced with the same external stimulus (e.g., AI awareness) vary owing to differences in personal traits. High achievement-motivated employees are more likely to appraise AI as a “controllable challenge,” thereby stimulating proactive behavior. Conversely, those with low achievement motivation perceived it as an “uncontrollable threat,” which would result in withdrawal behavior. Thus, the direction and strength of the impact of AI awareness on employee behavioral pathways vary significantly depending on the level of achievement motivation.

In summary, achievement motivation, as a critical internal personal resource, moderates how employees cognitively interpret changes brought about by AI awareness and shapes the direction of behavioral responses during the “cognitive appraisal–behavioral response” process. Therefore, this study proposes the following hypothesis:

Hypothesis 3a.

Achievement motivation positively moderates the relationship between AI awareness and proactive behavior, such that the higher the level of achievement motivation, the stronger the positive impact of AI awareness on proactive behavior.

Hypothesis 3b.

Achievement motivation negatively moderates the relationship between AI awareness and withdrawal behavior, such that the lower the level of achievement motivation, the stronger the positive impact of AI awareness on withdrawal behavior.

Building upon the above theoretical foundations, this study further argues that achievement motivation not only moderates the direct relationship between AI awareness and employee behavioral responses (i.e., proactive behavior and withdrawal behavior) but also shapes the strength of the indirect effects of AI awareness on employee innovation behavior through these behavioral pathways. In other words, this framework constitutes a moderated mediation model, in which the mediating effects of proactive and withdrawal behaviors are conditional upon the level of employees’ achievement motivation.

Specifically, when employees demonstrate a high level of achievement motivation, they are more likely to interpret AI-related changes as opportunities for challenge and self-improvement. This cognitive appraisal fosters higher levels of proactive engagement, thereby enhancing the positive mediating effect of proactive behavior in the link between AI awareness and innovation behavior. Conversely, employees with high achievement motivation are also more resilient in the face of potential threats or uncertainties brought about by AI. They are more capable of regulating negative emotional reactions and are thus less likely to engage in withdrawal behavior. Accordingly, achievement motivation is expected to weaken the negative mediating effect of withdrawal behavior on the relationship between AI awareness and innovation outcomes.

This moderated mediation perspective highlights the boundary condition under which the indirect effects of AI awareness vary, and it offers a more nuanced understanding of how individual-level motivational resources shape AI-induced behavioral and innovation outcomes. Accordingly, the following hypotheses are proposed:

Hypothesis 4a.

High achievement motivation positively moderates the mediating effect of proactive behavior in the relationship between AI awareness and employee innovation behavior.

Hypothesis 4b.

High achievement motivation negatively moderates the mediating effect of withdrawal behavior in the relationship between AI awareness and employee innovation behavior.

Based on the above theoretical hypotheses, this study constructs a moderated mediation model (see Figure 1) that explains how AI awareness influences employee innovation behavior through two distinct behavioral pathways. Specifically, AI awareness, as a perceived external stimulus, activates either proactive or withdrawal behavior depending on employees’ cognitive appraisals. These behavioral responses are further shaped by employees’ achievement motivation, which moderates the strength and direction of their reactions. The resulting behavioral outcomes provide insight into how individual perceptions of AI influence innovation behavior in high-tech work environments.

3. Methods

3.1. Sample and Data Collection

This study designed an online questionnaire using the Credamo platform and distributed the survey link to participants in December 2024 via an online survey method. A two-wave online survey design was employed to collect first-hand data from knowledge workers in Guangdong Province, China. The participants were primarily drawn from enterprises in knowledge-intensive industries, such as software development, smart manufacturing, internet platforms, and technology services.

The design and implementation of this study strictly followed the ethical guidelines for exempt studies established by the Research Ethics Committee of Gachon University, South Korea. The entire research process adhered to high ethical standards to ensure participants’ rights to informed consent, privacy, and voluntary participation. Specific measures included the following: First, the study collected only self-reported data regarding employees’ subjective perceptions and behavioral intentions without involving any form of physical or psychological intervention, thereby posing minimal risk to participants. Second, all data were collected anonymously and were accessible only to the research team, effectively minimizing the risk of disclosing sensitive information. Finally, prior to survey distribution, the research team fully informed all participants about the purpose and procedures of the study, clearly stating that their participation was entirely voluntary and that they could withdraw at any time without penalty. The confidentiality of all responses was explicitly guaranteed. Informed consent was obtained from all participants without any external pressure.

To reduce random errors commonly associated with cross-sectional studies, different variables were measured at separate time points [54]. In Time 1 (T1), participants provided demographic information along with data on AI awareness and achievement motivation. Four weeks later (Time 2, T2), the same participants reported their levels of proactive behavior, withdrawal behavior, and innovation behavior. At the T1 stage, an online questionnaire was distributed to 600 knowledge workers. After excluding invalid responses, 552 valid questionnaires were obtained. Four weeks later, the T2 survey was administered to participants who completed the T1 survey. After a second round of data screening, 413 valid paired responses were retained for analysis, resulting in an effective response rate of 68.83%.

Among the 413 valid respondents in this study, 260 were male (63.0%) and 153 were female (37.0%). In terms of age distribution, 158 respondents were aged 20–29 (38.3%), 172 were aged 30–39 (41.6%), 67 were aged 40–49 (16.2%), and 16 were aged 50 or above (3.9%). The majority of respondents held junior college or undergraduate degrees, totaling 362 individuals (87.7%). Regarding job positions, most respondents were ordinary employees, totaling 252 individuals (61.0%). In terms of tenure, more than half of the respondents had fewer than 5 years of work experience (68.5%), while 130 respondents (31.5%) had more than 5 years of work experience.

3.2. Measures

In this study, the initial scales were developed in English and then translated into Chinese using the back-translation method recommended by Brislin [55]. All measurement items were evaluated on a five-point scale. To ensure comparability across constructs, mean scores were calculated for each variable, including AI awareness, proactive behavior, withdrawal behavior, innovation behavior, and the two subdimensions of achievement motivation (pursuing success and avoiding failure). For the total achievement motivation score, items measuring avoidance motivation were reverse-coded prior to computing the overall average across all 30 items.

3.2.1. Innovation Behavior

Innovation behavior was gauged using six items developed by Hortinha et al. [56]. Example items are “I look for novel technological ideas by thinking outside the box” and “I base success on our ability to explore new technologies.”

3.2.2. AI Awareness

AI Awareness was gauged using four items developed by Brougham and Haar [57]. Example items are “I think my job could be replaced by AI” and “I am personally worried that what I do now in my job will be able to be replaced by AI.”

3.2.3. Proactive Behavior

Proactive behavior was gauged using nine items developed by Griffin et al. [58]. Example items are “I proactively make suggestions to improve the overall efficiency of the organization” and “I quickly adapt to new equipment, processes, or procedures in core tasks.”

3.2.4. Withdrawal Behavior

Withdrawal behavior was gauged using 12 items developed by Lehman and Simpson [59]. Example items are “I have thoughts about being absent from work” and “I sometimes think about being absent from work.”

3.2.5. Achievement Motivation

Performance-approach goal orientation was gauged using 30 items developed by Ye and Hagtvet [36]. Example items are “I enjoy persistently working on problems that I am not certain how to solve” and “I enjoy novel and challenging tasks, even if they involve taking risks.”

3.2.6. Control Variables

In the model analysis, several individual-level background variables were controlled to enhance the internal validity and explanatory power of the conclusions. Specifically, gender, age, and educational level were included as basic demographic variables, considering their potential systematic influence on employees’ innovative behavior. In addition, since the frequency of employee exposure to AI in their work and their occupational rank may significantly affect their level of AI awareness, this study incorporated tenure and position as control variables. This approach follows the recommendations of [17], who argued that tenure and position influence employees’ risk perception of technological change and their evaluation of related resources, which may confound the true effect of AI awareness on behavioral pathways. By controlling these variables, this study aims to identify the impact mechanism of AI awareness on employee behavior and innovation performance within a clearer causal framework.

3.3. Reliability and Validity Tests

SPSS 28.0 is used to test the Cronbach’s α coefficients of each variable. The results are shown in

Table 1. Reliability test results.

Variable	Number of Items	Cronbach’s α
AI Awareness	4	0.869
Achievement Motivation	30	0.928
Proactive Behavior	9	0.931
Withdrawal Behavior	12	0.948
Innovation Behavior	6	0.904

. The table shows that the Cronbach’s α of all variables is greater than 0.8, which indicates that the scales have high reliability, and the measurement questions are provided with good internal consistency.

3.4. Common Method Bias

Harman’s single-factor test was conducted to assess the presence of common method bias. The first-factor interpretation rate was only 21.182%, which is far lower than the standard requirement of 40%, indicating no serious systematic bias in the questionnaire results.

3.5. Confirmatory Factor Analysis and Correlation Analysis

In the validity test results of the questionnaire (

Table 2. Comparison of measurement models.

	χ2	df	χ2/df	RMSEA	SRMR	CFI	TLI
Reference	-	-	<3	<0.08	<0.08	>0.9	>0.9
Six-factor model	1992.090	1754	1.136	0.018	0.034	0.987	0.987
Five-factor model	6367.712	1759	3.620	0.080	0.148	0.754	0.744
Four-factor model	10,078.373	1763	5.717	0.107	0.197	0.555	0.539
Three-factor model	10,704.990	1766	6.062	0.111	0.204	0.522	0.505
Two-factor model	11,041.937	1768	6.245	0.113	0.193	0.504	0.487
Single-factor model	15,376.773	1769	8.692	0.137	0.233	0.273	0.247

Note: Six-factor model: EIB, AI, PB, WB, MPS, MAF; Five-factor model: EIB, AI, PB, WB, MPS + MAF; Four-factor model: EIB, AI, PB + WB, MPS + MAF; Three-factor model: EIB + AI, PB + WB, MPS + MAF; Two-factor model: EIB + AI + PB + WB, MPS + MAF; Single-factor model: EIB + AI + PB + WB + MPS + MAF.

), the theoretical model (six-factor model) fit index χ2/df = 1.136 < 3, RMSEA = 0.018 < 0.08, SRMR = 0.034 < 0.08, CFI = 0.987 > 0.9, and TLI = 0.987 > 0.9. All fit indices met the analysis requirements, and the model had strong structural validity. However, the fit indicators of the other competitive models did not meet the analysis requirements, and the fit parameters were significantly worse than those of the theoretical model. Therefore, the variables in the model exhibited strong discrimination; that is, the discriminant validity of the questionnaire met the requirements.

4. Results

4.1. Correlation Analysis

The results of the correlation analysis in

Table 3. Means, standard deviations, and correlations.

	AI	MPS	MAF	AM	PB	WB	EIB
AI	1
MPS	0.181 ***	1
MAF	−0.130 **	−0.129 **	1
AM	0.207 ***	0.753 ***	−0.749 ***	1
PB	0.621 ***	0.118 *	−0.168 **	0.190 ***	1
WB	0.312 ***	−0.104 *	0.148 *	−0.167 **	0.161 **	1
EIB	0.474 ***	0.138 **	−0.211 ***	0.232 ***	0.513 ***	−0.116 *	1
M	3.43	3.43	2.58	3.42	3.26	2.85	3.29
SD	0.97	0.96	0.95	0.72	1.20	1.01	0.96

Note: * p < 0.05; ** p < 0.01; *** p < 0.001. AI: AI Awareness; MPS: Motivation To Pursue Success; MAF: Motivation To Avoiding Failure; AM: Achievement Motivation; PB: Proactive Behavior; WB: Withdrawal Behavior; EIB: Employee Innovation Behavior.

show that there is a significant positive correlation between employees’ innovation behavior and AI awareness, proactive behavior, motivation to pursue success, and achievement motivation, while there is a significant negative correlation between employee innovation behavior and withdrawal behavior and motivation to avoid failure. There was also a significant correlation between the variables/dimensions, such as AI awareness, positive behavior, withdrawal behavior, motivation to pursue success, motivation to avoid failure, and achievement motivation, and the results of the variable correlation test met the prerequisites for the regression analysis.

4.2. Hypothesis Testing

The influencing factor analysis of AI awareness on employee innovative behavior showed that, among the control variables, gender (b = −0.266, p < 0.01), age (b = −0.245, p < 0.001), and tenure (b = −0.236, p < 0.001) had significant negative effects on employee innovative behavior. Specifically, male employees tend to exhibit higher levels of innovative behavior than female employees; however, innovative behavior significantly decreases with increasing age and work experience. In contrast, educational background has a significant positive effect on innovative employee behavior (b = 0.248, p < 0.001), indicating that employees with higher education levels tend to display more innovative behaviors.

In the test of the relationship between AI awareness and proactive and withdrawal behavior, AI awareness can significantly improve the level of proactive and withdrawal behavior among employees. Among them, the influence coefficient of AI awareness on proactive behavior is b = 0.652, p < 0.001, and the influence coefficient on withdrawal behavior is b = 0.449, p < 0.001, and the hypotheses H1a and H1b are valid. The results are shown in

Table 4. Influencing factors of employees’ innovation behavior.

	EIB
	M1	M2	M3
Constant	4.010 *** (0.371)	2.754 *** (0.355)	2.743 *** (0.345)
Gender	−0.266 ** (0.084)	−0.291 *** (0.076)	−0.231 ** (0.075)
Age	−0.245 *** (0.063)	−0.189 ** (0.057)	−0.141 * (0.057)
Education	0.248 *** (0.065)	0.198 ** (0.059)	0.115 (0.060)
Position	−0.030 (0.051)	−0.034 (0.046)	−0.037 (0.044)
Tenure	−0.236 *** (0.048)	−0.190 *** (0.044)	−0.133 ** (0.044)
AI		0.357 *** (0.038)	0.321 *** (0.047)
PB			0.149 ** (0.040)
WB			−0.137 ** (0.044)
R2	0.287	0.406	0.432
Adjust R2	0.278	0.398	0.421
F	32.712 ***	46.345 ***	38.485 ***

Note: * p < 0.05; ** p < 0.01; *** p < 0.001. AI: AI Awareness; PB: Proactive Behavior; WB: Withdrawal Behavior; EIB: Employee Innovation Behavior.

.

In Model 3, the mediating variables had a significant effect on the dependent variable (Table 4), whereas Models 5 and 7 show that the independent variable had a significant positive effect on both types of mediating variables, indicating the existence of a mediating effect in the model (

Table 5. Influencing factors of proactive and withdrawal behaviors.

	PB		WB
	M4	M5	M6	M7
Constant	2.818 *** (0.474)	0.523 (0.430)	2.073 *** (0.437)	0.490 (0.392)
Gender	−0.156 (0.107)	−0.202 * (0.092)	0.250 * (0.099)	0.218 * (0.084)
Age	−0.181 * (0.080)	−0.079 (0.069)	0.196 ** (0.074)	0.267 *** (0.063)
Education	0.478 *** (0.083)	0.385 *** (0.072)	−0.121 (0.077)	−0.185 ** (0.065)
Position	0.046 (0.065)	0.038 (0.056)	0.025 (0.060)	0.020 (0.051)
Tenure	−0.295 *** (0.062)	−0.209 *** (0.053)	0.127 * (0.057)	0.186 *** (0.048)
AI		0.652 *** (0.046)		0.449 *** (0.042)
R2	0.258	0.513	0.110	0.281
Adjust R2	0.249	0.506	0.099	0.270
F	28.272 ***	71.306 ***	10.035 ***	26.403 ***

Note: * p < 0.05; ** p < 0.01; *** p < 0.001. AI: AI Awareness; PB: Proactive Behavior; WB: Withdrawal Behavior.

).

The PROCESS macro was used to decompose and calculate the various effects within the model; the results are listed in

Table 6. Mediation effect test.

Effect Decomposition	Path	Effect	SE	LLCI	ULCI
Total Effect	AI→EIB	0.357	0.039	0.279	0.434
Direct Effect	AI→EIB	0.321	0.050	0.223	0.419
Mediation Effect	AI→PB→EIB	0.097	0.027	0.044	0.151
	AI→WB→EIB	−0.061	0.021	−0.102	−0.023

Note: AI: AI Awareness; PB: Proactive Behavior; WB: Withdrawal Behavior; EIB: Employee Innovation Behavior.

. According to the data, the total effect of AI awareness on innovative behavior is 0.357. The indirect effect of proactive behavior was 0.097, with a 95% confidence interval of [0.044, 0.151], which did not include 0. Therefore, the mediating effect of proactive behavior was established, and H2a was supported. The indirect effect of AI awareness on innovative behavior through withdrawal behavior was −0.061 with a 95% confidence interval of [−0.102, −0.023], which did not include 0. Hence, the mediating effect of withdrawal behavior was confirmed, and Hypothesis H2b was supported.

In the moderation effect test, the interaction term “AI × AM” (AI awareness × Achievement Motivation) has a significant positive effect on proactive behavior (b = 0.149, p < 0.05), indicating that as the level of achievement motivation increases, the positive impact of AI awareness on proactive behavior becomes stronger (see Figure 2), thus supporting Hypothesis H3a. Meanwhile, the interaction term “AI × AM” has a significant negative effect on withdrawal behavior (b = −0.259, p < 0.001), suggesting that the higher the level of achievement motivation, the weaker the positive impact of AI awareness on withdrawal behavior (see Figure 3), thereby supporting Hypothesis H3b. Detailed calculation results are presented in

Table 7. The moderating effect of achievement motivation.

	PB	WB
	M8	M9
Constant	2.699 *** (0.406)	2.171 *** (0.332)
Gender	−0.180 * (0.093)	0.154 (0.076)
Age	−0.076 (0.069)	0.259 *** (0.056)
Education	0.375 *** (0.071)	−0.157 * (0.058)
Position	0.037 (0.055)	0.006 (0.045)
Tenure	−0.194 *** (0.053)	0.151 ** (0.043)
AI	0.649 *** (0.047)	0.481 *** (0.038)
AM	0.025 (0.066)	−0.258 *** (0.054)
AI×AM	0.149 * (0.065)	−0.259 *** (0.053)
R2	0.520	0.341
Adjust R2	0.510	0.328
F	54.683	26.097

Note: * p < 0.05; ** p < 0.01; *** p < 0.001. AI:AI Awareness; AM: Achievement Motivation; PB: Proactive Behavior; WB: Withdrawal Behavior.

.

Further analysis was conducted to examine the differences in the sizes of the mediating effects under different levels of achievement motivation. The test results showed that when the level of the moderating variable (achievement motivation) was low, the mediating effect of the path “AI → PB → EIB” was only 0.081, while the mediating effect of the path “AI → WB → EIB” was −0.091. However, when achievement motivation was high, the mediating effect of the “AI → PB → EIB” path increased to 0.113, and the mediating effect of the “AI → WB → EIB” path decreased to −0.040. Further statistical analysis confirmed that the differences in the mediating effects between high and low achievement motivation levels were significant, indicating the presence of a significant moderated mediation effect, thereby supporting Hypotheses H4a and H4b. The detailed results are presented in

Table 8. Moderated mediation effect test.

Path	AM	Effect	SE	LLCI	ULCI
AI→PB→EIB	Low AM	0.081	0.023	0.037	0.129
	High AM	0.113	0.031	0.052	0.175
	Difference	0.032	0.016	0.006	0.067
AI→WB→EIB	Low AM	−0.091	0.032	−0.157	−0.033
	High AM	−0.040	0.014	−0.070	−0.015
	Difference	0.051	0.022	0.014	0.100

Note: AI: AI Awareness; AM: Achievement Motivation; PB: Proactive Behavior; WB: Withdrawal Behavior; EIB: Employee Innovation Behavior.

.

5. Discussion

This study focuses on knowledge-based employees in high-tech enterprises and focuses on how AI awareness influences employee innovation behavior through behavioral pathways. A theoretical model incorporating AI awareness, proactive/withdrawal behavior, and achievement motivation was constructed, and structural equation modeling was used to analyze the underlying mechanisms systematically. The main conclusions of this study are as follows:

First, it verifies that AI awareness influences employees’ innovation behavior through two distinct pathways, demonstrating significant bidirectional effects. AI awareness positively promotes employee innovation by stimulating proactive behavior [60]. However, it may also induce withdrawal behavior, thereby inhibiting innovation [61]. This indicates that employees’ subjective evaluations of AI play a critical role in their behavioral choices. Notably, although this study operationalized AI awareness as a unidimensional construct, the findings further underscore the theoretical necessity of distinguishing between different forms of AI awareness—such as perceived opportunity and perceived threat.

Second, proactive and withdrawal behaviors serve as positive and negative mediators, respectively, in the relationship between AI awareness and innovation behavior, forming a “dual-path competitive mechanism.” Proactive behavior, as a self-initiated, forward-looking, change-oriented tendency, enhances employees’ adaptability and innovation engagement in AI environments [31]. Conversely, withdrawal behavior reflects a psychological defense mechanism through which employees reduce their engagement owing to resource constraints or pressure overload [34]. This “dual-path mechanism” reveals the complexity and dynamism of behavioral choices when employees face AI-driven challenges.

Third, the study finds that achievement motivation played a critical moderating role in this mechanism. Employees with high achievement motivation are more likely to perceive AI as a challenge-based stressor and display more proactive behavior and a stronger innovation drive. By contrast, employees with low achievement motivation are more prone to perceiving AI as a source of uncertainty and threat, leading to more withdrawal behavior and diminished innovation intention. This finding highlights the importance of individual traits in adapting to AI technologies and provides theoretical support for organizations to implement differentiated incentives and capacity-building strategies.

5.1. Theoretical Implications

This study integrates perspectives from the “dual-stressor framework” and the “dual-path competition model” to systematically explore employees’ psychological mechanisms and behavioral responses to AI awareness. It constructs a dual-path model based on challenge-hindrance cognitive appraisal and reveals the mechanisms by which AI awareness facilitates or inhibits innovation behavior while introducing achievement motivation as a moderating variable to emphasize the boundary effects of individual differences in digital contexts. The primary theoretical contributions of this study are as follows:

First, this study constructs a system-oriented theoretical framework for AI awareness. Moving beyond the traditional view of AI as a single external technological stimulus, the study conceptualizes AI awareness as a subjective psychological input variable with dual attributes of challenge and hindrance, which suggests that AI awareness should be viewed as a subjective psychological construct with both challenge and hindrance attributes. This construct emphasizes that employees are not passive recipients of technological change but instead form differentiated psychological experiences and attitudinal tendencies through their own cognitive appraisals [62], resulting in heterogeneous behavioral responses. This theoretical perspective deepens our understanding of how employees perceive technology and provides a cognitive foundation for explaining behavioral differences in the face of AI.

Second, this study reveals the underlying mechanism through which AI awareness influences employee innovation behavior by identifying two opposing behavioral pathways: proactive and withdrawal behaviors. Proactive behavior reflects employees’ active adaptation and engagement in response to AI-driven change, while withdrawal behavior represents defensive and avoidant reactions under perceived technological pressure. These two pathways form a dual behavioral response pattern in the context of digital transformation, highlighting how employees’ subjective perceptions of AI affect innovation outcomes through differentiated behavioral trajectories. By introducing the dual-path competition model into AI-driven organizational behavior research, this study not only extends its applicability to emerging technological contexts but also enriches the theoretical connotation of the dual-path competition model in the behavioral domain [63], providing a structured analytical framework for understanding how employees respond to technological and motivational pressures.

Third, this study introduced achievement motivation as a moderating variable, highlighting its dual role in resource mobilization and cognitive transformation in AI-embedded organizational contexts. Specifically, individuals with high achievement motivation are more likely to view AI as an opportunity to enhance their performance and capabilities, thereby reinforcing proactive behaviors driven by challenge appraisals and suppressing negative behaviors arising from hindrance appraisals. This finding expands the theoretical applicability of achievement motivation in the digital age and responds to recent research on how psychological resources moderate the effects of technological shocks, emphasizing the critical role of psychological capital in organizational change processes.

Finally, this study draws on systems thinking at the theoretical level, aiming to holistically understand the interactions among AI awareness, behavioral pathways, and individual motivation within organizational contexts. These interactions are viewed as key psychological and behavioral mechanisms through which organizations respond to technological change. Although the research model is fundamentally linear and empirically tested using survey data, the analysis of these structured relationships contributes to understanding how individual differences influence innovation behavior across multiple levels of organizational functioning. This perspective offers a foundation for future multi-level and multi-variable theoretical development, with the potential to bridge micro-level cognitive mechanisms and macro-level organizational adaptability, thereby expanding the theoretical boundaries of organizational behavior in the context of digital transformation.

5.2. Practical Implications

This study provides several actionable managerial insights grounded in systems thinking, aimed at helping organizations optimize human resource strategies and guide employee behavioral patterns during AI-driven transformation. These recommendations contribute to enhancing the adaptability of human–AI collaboration and support dynamic regulation within increasingly complex organizational environments.

First, organizations should move beyond the traditional linear mindset that treats AI as an isolated technological tool and instead adopt a more integrated management perspective that regards AI as a core element deeply embedded in organizational operations. As key agents of behavioral transformation, employees’ perceptions and responses to AI play a critical role in shaping organizational innovation capacity and operational efficiency. To manage this effectively, organizations should establish actionable perception and feedback mechanisms, such as: regularly conducting AI awareness assessment surveys to understand employees’ comprehension and acceptance of AI tools; deploying digital tools to monitor psychological stress and emotional well-being, such as psychological evaluation systems or anonymous emotional surveys; and building employee feedback platforms to collect real-time concerns and suggestions related to AI usage. These tools enable managers to dynamically track employee conditions, identify potential adaptation risks, and implement personalized interventions. In doing so, organizations can effectively implement differentiated and tiered human resource management strategies, thereby enhancing collaborative efficiency during the AI-driven transformation process.

Second, to enable more personalized and system-oriented managerial interventions, organizations can adopt a cross-classification strategy based on employees’ stress appraisal types (challenge stressors vs. hindrance stressors) and levels of achievement motivation (high vs. low). For example, employees experiencing high challenge stress and high achievement motivation can be assigned to “AI transformation pilot teams,” provided with greater autonomy and resources, and stimulated through goal-oriented tasks to foster proactivity and innovation. Those with high challenge stress but low achievement motivation may benefit from regular developmental feedback and targeted incentive programs to strengthen engagement and goal commitment. For employees under high hindrance stress and high achievement motivation, organizations can offer cognitive reframing interventions and stress adaptation strategies to help convert pressure into constructive behavioral responses. Meanwhile, individuals with high hindrance stress and low achievement motivation should be prioritized for emotional counseling and basic AI skills training to reduce anxiety and gradually build adaptive capacity. By applying this “stress × motivation” quadrant-based intervention logic, organizations can more effectively tailor their HR strategies to diverse employee profiles, thereby improving adaptive behavior and enhancing organizational resilience during AI-driven transformations.

Finally, at the organizational level, performance incentive mechanisms that emphasize “human–AI collaboration capability” should be designed to establish a positive linkage between employee behavioral responses and the achievement of organizational goals. By introducing growth-oriented evaluation metrics, innovation-based incentive point systems, and AI collaboration capability rating models, organizations can construct a closed-loop feedback structure between the behavioral subsystem and the strategic subsystem. This enables a positive coupling among employee cognition, behavior, and organizational adaptability, thereby enhancing the overall stability and resilience of the organizational system during AI-driven transformation.

5.3. Limitations and Future Research

First, this study adopts a cross-sectional design, which, despite using time-lagged measurements to partially reduce common method bias, remains limited in capturing the dynamic cognitive evolution of employees’ AI awareness and the feedback mechanisms operating within the organizational system. Systems theory emphasizes the temporal and recursive nature of behavioral feedback, and static designs are insufficient for revealing causal chains and phased progression among variables. Therefore, future research should consider employing longitudinal tracking, scenario-based experiments, or system dynamics modeling to simulate the temporal trajectory linking AI cognition, behavioral feedback, and performance outcomes. Such approaches would offer a more comprehensive understanding of the dynamic coupling mechanisms within organizational systems.

Second, this study’s sample is drawn exclusively from knowledge workers in China’s high-tech sector. Although this group is particularly responsive to AI-driven changes, the findings may not generalize across different cultural or industrial contexts. For example, employees in individualistic cultures may exhibit different cognitive appraisal patterns, motivational orientations, and behavioral reactions to AI stimuli compared to those in collectivist cultures like China. Similarly, industries such as manufacturing, service, or public administration may involve different task characteristics, technological familiarity, and perceptions of AI-induced role disruption. Future research is therefore encouraged to replicate and extend the proposed model across diverse national, cultural, and industrial settings to examine its cross-contextual robustness and theoretical boundary conditions.

Third, this study adopted the four-item scale developed by Brougham and Haar [57] to measure AI awareness as a unidimensional construct, primarily focusing on employees’ perceived risk of being replaced by AI. As a result, the current research does not distinguish between different cognitive dimensions such as “perceived threat” and “perceived opportunity,” which represents a theoretical simplification. Future studies are encouraged to develop and validate a multidimensional scale of AI awareness that incorporates both threat and opportunity appraisals, thereby allowing a more nuanced examination of the heterogeneous behavioral pathways driven by different types of perceptions. Additionally, subsequent research may incorporate multilevel variables such as team climate and organizational support to build a more ecosystem-based model of AI adaptation.

Finally, the present study preliminarily constructs a “cognition-driven behavioral mechanism,” which begins with AI awareness, is mediated by behavioral pathways, and centers on motivational regulation. This framework reveals how individuals make behavioral choices under uncertain, future-oriented knowledge contexts. Future research may further incorporate the Knowledge Field Theory [64] and Knowledge Dynamics Theory [65] proposed by Bratianu and Bejinaru to deepen the understanding of employees’ cognitive transformation processes and the flow of knowledge energy within organizations. Such integration could help expand the theoretical boundaries of nonlinear interaction mechanisms between AI awareness and employee behavior. By adopting a knowledge management perspective, future studies may enrich the theoretical connotation of this mechanism and offer more systematic and dynamic insights into how employees’ cognition evolves and how behavioral feedback functions during technological transformation.

6. Conclusions

This study confirms that AI awareness can both stimulate and inhibit employees’ innovation behavior, depending on whether it triggers proactive behavior or withdrawal behavior. Furthermore, achievement motivation plays a critical moderating role in this process, amplifying positive behavioral responses while mitigating negative ones. Based on these findings, this study constructs and empirically validates a “dual-path mediation + unidirectional moderation” model to explain how AI awareness influences innovation behavior through behavioral mechanisms. Although the model focuses on the individual level, the collective behavioral changes of employees may, over time, accumulate and evolve into systemic organizational adaptation patterns, forming a micro-level behavioral foundation for coping with technological transformation. Therefore, this research not only deepens our understanding of employee behavioral mechanisms in response to emerging technologies but also extends the application of Cognitive Appraisal Theory within the context of digital transformation. The findings provide valuable theoretical insights and practical guidance for organizations navigating AI-driven change.