1. Estimate the levels of GenAI anxiety and learning motivation among EFL gifted learners.

2. Test whether gifted behavior and GenAI anxiety predict learning motivation.

3. Examine whether GenAI anxiety mediates the gifted behavior → learning motivation link.

4. Assess whether gender moderates structural paths among gifted behavior, GenAI anxiety, and learning motivation.

EFL students have heterogeneous motivation, strategy profiles, and informal digital learning profiles -differences that are usually magnified among gifted students. Person-centered research designates different clusters of low quantity and high quantity motivation with gendered distributions (Li et al., 2023), three profiles of self-regulated learning influenced by self-efficacy and extrinsic motivation (Chen et al., 2023), and various digital-informal learning profiles that are associated with increased L2 motivation, pleasure, and grit (Lee and Xie, 2023). The interest-based, hedonistic involvement seems to be typical of successful learners, which makes programmatic perspectives of input difficult (Tsang, 2023). In the area of gifted education, there is also a combination of performance gains and unique cognitive-affective profiles capable of either hastening or challenging the development of a second language (L2) (Heidari, 2025). Gifted-specific reviews and AI reviews highlight the potential of personalization, yet they claim to need evidence-based designs that are sensitive to the socio-emotional needs of gifted learners (García-López et al., 2025). There is also complementary evidence indicating that motivation and grit are part of self-managed learning and success, as well as certain gender-specific mechanisms (Guo et al., 2023).

EFL studies in Saudi Arabia have revealed significant achievements that are in tandem with the national transformation agenda. The Saudi Arabian economic diversification plan, Vision 2030, has an intricate causal relationship with EFL learning development; however, the directional effect is yet to be seen (Al-Mwzaiji and Muhammad, 2023). According to the latest research, the Saudi EFL learners have incredibly positive attitudes to AI-enabled language learning technologies, and the integration of AI-powered tools into the learning process is strongly supported by the learners (Jamshed et al., 2024). Nevertheless, even supportive attitudes do not address the implementation problems (Jamshed et al., 2024). It is statistically found that the level of study has a significant impact on the perception of learners, whereas gender, parental employment, residential background, and parental education do not produce significant differences (Jamshed et al., 2024). Within the broader context of gifted education, the aforementioned state of Saudi Arabia has been assessed through the Educational and Learning Capital Model, which has identified the strengths and aspects that need improvement in the delivery of gifted students (Alfaiz et al., 2022).

MALL/CALL has positive relations with EFL motivation in the form of attractiveness, usefulness, and authenticity (Wei, 2022). The level of motivation is a strong predictor of engagement (Wang, 2022), correlates with grit and intrinsic orientation (Liu, 2022), and is associated with resilience (Zhang, 2022). At the level of development, the intensity of motivation depends on grade levels and contexts (Cheng, 2024). GenAI studies have shown lower language anxiety and improved proficiency (Chen et al., 2025; Wang, 2025), as well as higher autonomy and ESP writing proficiency gains (Mohamed et al., 2025), and motivational/affective advantages (Wang et al., 2025). Nevertheless, the anxiety associated with GenAI has the potential to moderate adoption and use, in interaction with AI literacy, self-efficacy, and classroom feelings (Wang et al., 2024; Cengiz and Peker, 2025; Zhang et al., 2025). According to meta-reviews, AI-assisted learning has a dual affective profile: it is both motivating and stress-inducing (Zhang, 2025). Motivation and engagement are enhanced by task designs (e.g., oral reading with AI) and learner characteristics (e.g., gender, level, and proficiency) (Alfaleh et al., 2025).

The gendered trends in AI adoption are always present: women tend to express greater GenAI anxiety and more wary attitudes than men (Russo et al., 2025), but can be even more successful in AI-assisted speaking (Fauzi et al., 2025). Anxiety may serve to reduce attitudinal differences due to gender differences leveling at high anxiety rates (Russo et al., 2025). Students in the classroom using AI technology experience a range of emotions, alternating between excitement and anxiety (Xin and Derakhshan, 2025). Regarding motivation, females tend to show more overall, intrinsic, and integrative motivation (Alghamdi et al., 2023; Tanaka, 2023), whereas males tend to show more controlled motivation (Li et al., 2023). There might be an achievement pattern that favors females even when grit/mindset are the same between genders, and the prediction differs based on gender (Guo et al., 2023). These patterns have been explained through social cognitive explanations, which suggest that they result from gendered self-efficacy and stereotype-informed experiences (Kutuk, 2025; Liu et al., 2024), despite some cases of null moderation (Shen and Bai, 2024).

The implication of a broad literature review demonstrates a significant and complex research gap. Despite the already confirmed fact that significant part of research has been carried out on gifted education (Heidari, 2025; García-López et al., 2025), EFL motivation (Wei, 2022; Liu, 2022), technology anxiety (Wang et al., 2024; Zhang, 2025), and gender differences as independent variables (Russo et al., 2025; Li et al., 2023), a combination of these lines of research has not The literature regarding GenAI in EFL has mainly focused on general populations of learners, which document the effect of such a technology on proficiency and the overall types of anxiety (Chen et al., 2025; Wang, 2025). Conversely, the studies of gifted EFL learners focus more on cognitive, motivational, or teaching aspects, and rarely address the emerging AI-based technologies (Heidari, 2025; García-López et al., 2025).

In addition, even though it is sufficiently documented that gender could moderate the usage of technologies and motivation (Russo et al., 2025; Li et al., 2023), there is a shortage of empirical data regarding how gender can be factored into the models that simultaneously consider giftedness, GenAI anxiety, and academic motivation, particularly when viewed through the prism of Control-Value Theory and Social Cognitive Theory.

To address these gaps, the paper will not contradict previous studies but will build on them. Using the longstanding scales of gifted behavior, GenAI anxiety, and academic motivation, the study no longer relies on the simple main effects designs but initiates a theoretically constructed and empirically testable model of relationships where relationships are estimated as solutions in terms of the mediating and moderating effects of gender. This way, the study would contribute to a more subtle perspective of the psychological mechanisms underlying the academic motivation of gifted EFL students in fast-paced learning environments, which implies the use of GenAI. It is anticipated that the findings will provide both theoretical and pedagogical implications that can be used to develop supportive instructional approaches to capitalize on the opportunities of gifted students through the utilization of GenAI in a manner that enhances and does not limit their potential, without being insensitive to other individual differences, such as gender.

This study adopted a quantitative correlational research design, implemented through a cross-sectional survey. Structural equation modeling (SEM) was employed to examine the predictive and mediating relationships among gifted behavior, GenAI anxiety, and learning motivation, as well as to assess gender-based moderation. This design was chosen because it allows for the simultaneous estimation of complex relationships among latent variables, which is essential for testing the hypothesized theoretical model.

Students attending the two Saudi Arabian universities who are studying English courses were selected through a purposive sampling technique. A total of 304 undergraduates completed the final sample, meeting the set of inclusion criteria (male = 71.7% and female = 28.3%). The definition of giftedness was based on the personal international standards of giftedness (Pfeiffer and Jarosewich, 2003) and the recent research on gifted students in technology-based learning settings (Ayik and Gül, 2025; Hulsey et al., 2023).

The institutions of higher learning investigated in this analysis were not selected based on convenience, but rather on methodology and context. The situation in Saudi Arabia presents a suitable context, as this country is actively transitioning to a digital environment, thoroughly implementing educational technologies, and English teaching is a dominant presence in higher education. The chosen institutions have similar EFL curricula and technological systems, which facilitates the minimization of the institutional variation and contributes to the internal validity. Besides that, the personal academic interaction of the researchers with these institutions enabled easy access to participants and the collection of data as well as compliance with ethical and procedural guidelines.

In order to qualify, students had to have official records of giftedness from institutional gifted programs or university-level gifted support services, which is typical in research on giftedness. They were also required to be enrolled in higher education, to be non-native speakers of English and to exhibit a consistently high success in EFL courses, in the form of a GPA ranking in the 10% of their cohort - a method that was previously applied to identify gifted language learners (Harris et al., 2009; Karami and Izadpanah, 2022; Alkhalifah and Alarfaj, 2023).

The short version of the Gifted Rating Scales (GRS) (Pfeiffer and Jarosewich, 2003) was also conducted on all participants to confirm that they have gifted behavioral traits of fast learning and creative problem solving, which are also prioritized in the recent research on AI-based gifted learning (Walters, 2024; Mohamed et al., 2025). Individuals with a score that exceeded the giftedness cut-off of the GRS were retained only, and it provided a chance of strong psychometric validation and institutional records. Such a two-screening process enhanced the internal validity of the sample.

The recruitment of participants was facilitated through the gifted support and honors programs in both universities, which checked the eligibility of the participants and also provided ethical access to the target population.

Gifted EFL learners in this research paper are considered to be those students who demonstrate high potential or achievement in English as a Foreign Language expressed not only through highly developed language skills but also in high motivation, self-regulation, and interest in complicated learning activities. Recent studies think of giftedness in EFL as a multidimensional and context-sensitive phenomenon and is determined by learner identity and sensitivity to enriched, technology-based learning conditions (Trân and Hoàng, 2024; Oto-Millera et al., 2025; Yang et al., 2025). Based on this, validated measures of gifted learning behaviors in EFL contexts are used in the current study to determine gifted EFL learners.

Prior to conducting the study, ethical approval was obtained from the Ethical Approval Committee, and all procedures were conducted in accordance with the approved ethical guidelines. Informed consent was obtained online. At the beginning of the study, participants received a digital consent form stating the study’s purpose, confidentiality, and voluntary participation. Only after clicking the “I agree to participate” button were participants allowed to continue with the study. The eligible students then completed an anonymous online survey, which consisted of demographic questions, the GRS, the GenAI Anxiety Scale, and the Academic Motivation Scale. All information was gathered on a secure site and stored in encrypted, password-protected files that are accessible only to the research team.

The research was conducted in two Saudi Arabian universities that teach English as a foreign language and have organized systems of recognizing and providing support to gifted students that are in practice. The Saudi higher education system has developed gifted education in the past few years with clear policies, institutional programs, and culturally based practices aiming at identifying and developing exceptional students (Aboud, 2025). Such universities also offer special programs, including enrichment programs and honors programs, which help gifted students in their academic progress, which are broader national trends in gifted care and education technology (Tawil and Tarawneh, 2026). Among these systems, gifted students are typically defined as students with high cognitive abilities, academic success, and learning requirements, which are also reported in studies of Saudi gifted students (Alhossein et al., 2025; Li et al., 2023).

Simultaneously, Saudi EFL classrooms are gradually becoming more digitalized, incorporating AI-based learning resources, which align with the worldwide trend of using technology in language learning (Alfaleh et al., 2025). These technologies, including AI writing assistants or adaptive feedback systems, are promising to provide new possibilities of personal assistance, but they also introduce an aspect of affective factors, such as diverse levels of confidence, uncertainty, or anxiety in the relationship with generative AI (Cengiz and Peker, 2025; Wang, 2025). These emotional dynamics can be particularly relevant to gifted students, who appreciate the level of precision, autonomy, and moral clarity in their learning settings (García-López et al., 2025).

Gender plays a significant role in the experience of these AI-assisted EFL environments. Studies have demonstrated that male and female students may differ in terms of motivation, emotional control, and responses to new technologies, particularly in solving AI-intensive learning tasks (Alfaleh et al., 2025; Russo et al., 2025). These trends render the Saudi gifted EFL setting to be a fruitful place to investigate the relationship between gifted behavioral traits and GenAI anxiety and learning motivation, and how such relationships might vary by gender, within technologically changing university language courses.

The survey was a multi-section online tool comprised of four valid instruments. The initial part was a Demographic Information Questionnaire created in the frame of the current study, and the information about age, gender, grade level, years of studying English, prior experience with AI tools (e.g., ChatGPT, Grammarly), and perceived English proficiency were collected. The second part used the Gifted Rating Scales (GRS) (Pfeiffer and Jarosewich, 2003), a 10-item scale to measure cognitive and behavioral attributes of giftedness. Statements of this nature, like: I learn new things quickly; I solve problems creatively; were rated in a five-point Likert scale (1 = Not at all like me; 5 = Very much like me) with high scores indicating increased gifted behavioral tendencies. The scale has shown a high level of reliability and validity in previous studies.

The third scale was the Artificial Intelligence Anxiety Scale (AIAS), which is an eight-item instrument created on the basis of recent literature (Chen et al., 2025; Liu, 2025; Ramadini and Pratiwi, 2025; Salimi et al., 2025) to measure anxiety of students about the application of AI in schools. Questions such as I feel nervous using AI applications like ChatGPT and I am afraid that AI will replace human teachers were considered on a five-point Likert scale (1 = Strongly Disagree; 5 = Strongly Agree), with a higher score being a sign of higher GenAI anxiety. The scale was developed to be context-specific to the target population.

The last section involved an adapted 9-item version of the Academic Motivation Scale (AMS-C28) (Vallerand et al., 1992) that assessed intrinsic motivation (3 items), extrinsic motivation (3 items), as well as amotivation (3 items). One item was added to the amotivation construct, so they became (3 items). Some of the sample items were as follows: I study English because I love to learn new things, I study English to achieve good grades, and I do not see it necessary to study English. Responses were measured on a seven-point Likert scale (1 = Not at all true; 7 = Very true), where higher scores indicate a stronger level of each motivational dimension. The Academic Motivation Scale (AMS) is widely used and has demonstrated good psychometric properties.

Before full data collection, all instruments underwent a two-stage validation procedure. First, an expert review involving three specialists in language assessment and educational psychology was conducted to examine the relevance, clarity, and construct representation of the items. The resulting item-content validity indices (I-CVIs) ranged from 0.88 to 1.00, indicating strong agreement among reviewers regarding the adequacy of each item. This was followed by a cognitive interview with ten EFL students to evaluate item comprehension and response processes. Minor wording adjustments were made based on participants’ feedback to enhance clarity and reduce ambiguity.

Because the instruments had been previously validated by the original authors, yielding the present factor structures, we conducted confirmatory factor analysis (CFA) to verify their structural adequacy in the present context. The CFAs demonstrated acceptable model fit across the three scales (Table 1), with fit indices falling within recommended thresholds (χ²/df > 1.5, RMSEA < 0.08, CFI > 0.9, TLI > 0.9, SRMR < 0.08). Standardized loadings were all substantial, ranging from 0.72 to 0.88 for Gifted Behavior, 0.74–0.88 for GenAI Anxiety, and 0.73–0.86 across the three motivation dimensions. Standardized loadings were all substantial, ranging from 0.72 to 0.88 for GBS, 0.74–0.88 for AIAS, and 0.73–0.86 across the three dimensions of the AIMS (Figure 3).

Evidence of convergent validity was supported by AVE values exceeding the.50 benchmark for all constructs (Table 2). Composite reliability (CR) and Cronbach’s alpha also met recommended thresholds, indicating strong internal consistency. Discriminant validity was confirmed through AVE–SIC comparisons, as each construct’s AVE exceeded the squared correlations with other constructs. These validation metrics demonstrate the validity and reliability of the instruments, thus indicating their suitability for the study.

The data analysis was conducted in several steps. Prior to the testing of the structural relationships, a series of preliminary diagnostics was performed in order to determine whether the dataset was suitable to multivariate modeling. The test of normality was also done using descriptive statistics (skewness and kurtosis) and through visual analysis using histograms with overlaid normal curves. The measure invariance between the genders was assessed through an item-level mean comparison procedure, which relied on the Smallest Effect Size of Interest (SESOI). The mean of the items for male and female learners was plotted against each other in view of the agreement and divergence in each construct. Scatterplots were to be employed to test the equality line alignment between the means of items, and the bar graphs were employed to reflect the difference between the means of the males and females, and the thresholds of SESOI were set to the practical equivalence of the results was determined to be within the range of −0.20 and +0.20. Differing items within this range of SESOI were regarded as being the same in both genders. Full collinearity variance inflation factors (VIFs) and a Common Latent Factor (CLF) procedure were used to measure the possibility of common method bias. These relative tests enabled us to assess the stability of the model estimates with respect to method variation.

After determining the quality of measurement, the structural relationships were tested through structural equation modeling (SEM) in AMOS. Analysis was based on estimating the direct effects of gifted behavior and GenAI anxiety, with the control variable of prior experience and English proficiency. To determine the mediation effects, the bias corrected bootstrapping parametric, with 5,000 resamples, was applied in order to get strong confidence intervals of the indirect effects. Multi-group SEM was used to test whether gender moderated the structural relations in the model, where the path coefficients were compared across male and female groups through equality constrained chi-square difference tests. The maximum likelihood method was used to estimate all the parameters, and unstandardized and standardized coefficients were calculated to facilitate the interpretability and comparability across the models.

As can be seen through the visual inspection of the histograms superimposed with normal curves, the distributions of the data was close to normality. The means of all of them were between 3.57 and 3.70 with standard deviations of 0.29–0.33. The values of skew were low (0.03–0.05), and kurtosis ranged from −0.40 to 0.48, indicating that there were no extreme violations of the normal distribution assumptions (Figure 4). The comparison of the item-level mean plots showed that male and female learners exhibited well-matched means on the items of the GenAI anxiety, the gifted behavior, and motivation. In the scatterplots, the means of the items were tightly clustered around the line of equality (45-degree line), indicating consistency in the functioning of items across the groups. It was also shown in the mean difference plots that all the differences were within the predetermined limits of the SESOI, that is, within the range of 0.20, indicating that the constructs used to measure the two genders were equivalent in measurement (Figure 5).

The VIFs of all the full-collinearity factors were within the range of 1.009–1.021; much lower than the maximum 3.3 (Table 3). The values indicate that there is no problematic multicollinearity, and the likelihood of common method variance inflating parameter estimates is low. The CLF analysis revealed that there was insignificant inflation of standardized factor loadings (Table 4). The difference between the baseline and CLF-adjusted loadings increased by +3.5–4.8% across all items, which is much low than the levels that are usually regarded as indicative of substantive common method bias. These findings also demonstrate that the method variance was not a significant distortion of the measurement structure.

The structural model revealed several significant direct effects (Figure 6 and Table 5). Gifted behavior negatively predicted GenAI anxiety (β = −0.26, p < 0.001) but demonstrated a positive and significant direct effect on learning motivation (β = 0.29, p < 0.001). In addition, GenAI anxiety negatively predicted learning motivation (β = −0.34, p < 0.001). Both covariates contributed significantly to the model. Prior experience negatively predicted GenAI anxiety (β = −0.19, p < 0.001) and positively predicted motivation (β = 0.17, p = 0.002). English proficiency showed a similar pattern, negatively predicting GenAI anxiety (β = −0.10, p = 0.028) and positively predicting motivation (β = 0.21, p < 0.001). The mediation analysis indicated a significant indirect effect of gifted behavior on motivation through GenAI anxiety (β = 0.09, p < 0.001). This confirms that the relationship between gifted behavior and motivation is partially conveyed through variations in GenAI anxiety.

The multi-group SEM showed that several structural paths differed across gender groups. As shown in Tables 6, 7, the direct effect of gifted behavior on GenAI anxiety was negative and statistically significant for both males (β = −0.24, p < 0.001) and females (β = −0.26, p < .001), and the equality-constrained test indicated no meaningful gender difference in this path, as indicated in Table 2 (ΔB = 0.02, z = 0.20, p = 0.840). In contrast, the path from gifted behavior to motivation varied significantly by gender. This association was positive for both groups but stronger among females (β = 0.42, p < 0.001) than males (β = 0.21, p = 0.004), with the difference reaching statistical significance (ΔB = −-0.24, z = −2.26, p = 0.024). A similar gender-differentiated pattern was observed for the effect of GenAI anxiety on motivation. Although GenAI anxiety negatively predicted motivation for both males (β = −0.26, p < 0.001) and females (β = −0.47, p < 0.001), the equality-constrained test revealed that the magnitude of this association was significantly stronger among females (ΔB = 0.22, z = 2.07, p = 0.038). Overall, the multi-group comparison indicates that gender moderates the associations between gifted behavior, GenAI anxiety, and motivation, but not the association between gifted behavior and GenAI anxiety.

The descriptive findings indicated that gifted EFL learners indicated relatively high motivation and moderate levels of GenAI-related anxiety. This seeming contradiction aligns with the growing body of literature, which observes that high-performing students tend to have mixed feelings toward AI solutions, exhibiting both excitement and caution (Xin and Derakhshan, 2025; Wang, 2025). It is possible that gifted students, due to their cognitive sensitivity, mastery orientation, and awareness of academic expectations, are motivated, at the same time, to increase anxiety about new and potentially disruptive tools.

This emotional multifacetedness highlights the fact that motivation is not identical to comfort. Among gifted students who value intellectual independence and accuracy, GenAI generates relevant issues regarding accuracy, ethical reliability, and authorship (García-López et al., 2025). As a result, such students are left in a unique state of emotional subculture in AI-mediated learning; they are compelled to use AI to improve their performance, but they are highly aware of its cognitive and moral consequences.

Gifted behavior showed a strong negative association with GenAI anxiety and a strong positive association with learning motivation. Such a pattern suggests that attributes such as higher reasoning, problem-solving flexibility, and rapid knowledge acquisition are associated with lower reported levels of technology-related anxiety. This is conceptually consistent with expectancy-value and self-efficacy models, in which perceived competence is linked to less negative emotional reactions to challenging tasks or tools (Liu, 2022; Tanaka, 2023).

Furthermore, the observed positive association between gifted behavior and motivation is consistent with existing findings in gifted education, which suggest that high cognitive ability is often correlated with intrinsic motivation, persistence, and mastery goals (Cheng, 2024; Li et al., 2023). These traditional motivational dynamics appear to be present in GenAI-mediated learning environments, conceptualized here as involving a broad category of GenAI tools, indicating that cognitive giftedness co-occurs with both lower anxiety and higher engagement in such contexts.

The mediation analysis provided a model to statistically unpack the associations described above. GenAI anxiety was found to be a significant mediator in the association between gifted behavior and learning motivation. It indicates that the observed association between gifted characteristics and motivation in our model was represented by a direct cognitive path and an indirect path statistically accounted for by emotional reactions to AI.

To further elucidate this statistical relationship, we rely on Control-Value Theory (Pekrun, 2006), which suggests that achievement-related emotions, such as anxiety, are linked to learners’ appraisals of a task. We propose CVT as a useful interpretive framework because the core attributes of gifted behavior measured in this study (e.g., problem-solving flexibility, rapid knowledge acquisition) are conceptually aligned with the antecedents of control appraisals, such as high self-efficacy and perceived competence.

To be more precise, anxiety is linked to the assessment of a lack of control over a procedure or the outcome of any task, and a high value placed on success (i.e., importance). Gifted behavior, in this regard, may be associated with initial perceptions of increased perceived control (e.g., self-efficacy in using AI tools) and favorable task value (e.g., considering GenAI to have intellectual value). Such appraisals, in their turn, are theorized to be linked to lower anxiety. GenAI anxiety, therefore, functioned as a key statistical mediator that accounted for the association between these cognitive-affective appraisals and motivational outcomes within the tested model.

Overall, the partial mediation analysis suggests that gifted traits are associated with higher motivation, both directly and indirectly through their relationship with lower levels of GenAI anxiety. This aligns with affective mediation theories in which emotions are posited to act as filters to cognitive involvement and academic persistence (Alrabai, 2025; Zheng, 2024). Given that gifted learners are often characterized by heightened sensitivities, future research may examine whether interventions that strengthen control appraisals are associated with more adaptive outcomes (e.g., by developing AI self-efficacy) and positively influence value appraisals (e.g., authentic learning with AI). Beyond comfort alone, appraisal processes may be associated with conditions that support the motivational potential of gifted learners in GenAI-enhanced environments (Wang et al., 2025).

Gender significantly moderated key relationships within the model. Although no mean differences in gifted behavior or GenAI anxiety were observed, the patterns of association with motivation differed. For female learners, gifted behavior was a stronger predictor of motivation, while GenAI anxiety exhibited a more pronounced negative association with motivational outcomes.

These findings reflect broader gendered patterns in academic and technological engagement. Female students often demonstrate strong self-regulation and commitment but can be more susceptible to affective dampening from technological insecurity or perceived threat (Russo et al., 2025; Liu et al., 2024). Thus, gifted female learners may experience heightened motivational returns from their cognitive strengths, yet also greater vulnerability to motivation loss if GenAI anxiety is elevated.

This duality highlights that gender functions not merely as a demographic variable, but as a psychological lens shaping differential engagement in GenAI-enhanced environments, consistent with social-cognitive theories of gendered expectations and self-perception (Kutuk, 2025).

Theoretically, the results support the new model that places AI-related emotions like GenAI anxiety in the middle of determinants that affect the motivation of gifted learners. Through its clear mediating position, the study can be used as a contribution to affective-cognitive theories of technology-enhanced learning, where it shows that giftedness does not necessarily lead to the best possible engagement unless the meaning of emotional barriers is addressed. The gender-moderation also builds on the motivational and emotional theories by exposing disparate zones by which both male and female gifted learners tend to follow AI-supported learning environments.

Pedagogically, the findings demonstrate the necessity of specific interventions that lessen the anxiety about AI, especially in gifted female students who demonstrate an increased emotional vulnerability. Teachers should consider incorporating explicit AI literacy teaching, scaffolding learning experiences that develop competence and confidence, and providing opportunities to experiment safely with AI tools. In addition, the creators of the curriculum must incorporate reflective and metacognitive tasks that can assist gifted students in working through the drawbacks and benefits of GenAI systems. Schools and universities must ensure that the AI-assisted learning environment is transparent, morally framed, and provided with human guidance to avoid overreliance and alleviate anxiety.

In general, teachers are to develop AI-mediated EFL environments, which are both supportive of emotions and cognitively challenging and sensitive to the diverse profiles of gifted students.

Multiple limitations are to be taken into consideration. First, the research design was cross-sectional, which does not allow for making causal conclusions and does not provide the possibility of observing changes in anxiety or motivation in the long term. Second, self-report measures are prone to social desirability bias and may not always be able to sufficiently apprehend the complexity of learners’ affective states. Third, the research was limited to two universities, and this might limit the generalization to other regions or education systems. Fourth, there was a significant gender imbalance in the sample (71.7% male). Although multi-group structural equation modeling was used to test moderation effects, this imbalance limits the stability and generalizability of the gender-specific associations observed in the model. These moderated relationships need to be confirmed by future studies that aim to find more balanced samples.

Fifth, the research included Generative AI as a generic term. Various GenAI applications (e.g., text generators, chatbots, coding assistants) can cause different affective and motivational reactions. The results are therefore generalized perceptions, and future research should be conducted to explore the variation of anxiety and motivation among specific types of AI tools and uses.

Future research should employ longitudinal or experimental designs to investigate the changes in GenAI anxiety and motivation over extended use of AI tools. It is advisable to use broader samples in different cultural, linguistic, and educational settings to increase the degree of generalizability. The effects of various AI tools on affective and motivational variables should also be examined by researchers in their unique ways: writing assistants, speech analyzers, adaptive feedback systems, and so forth. Lastly, future research should incorporate qualitative studies to gain a deeper understanding of how gifted learners perceive, internalize, and negotiate the presence of AI in their language-learning experiences.