A cross-sectional survey study published in Discover Artificial Intelligence in November 2026 found that self-efficacy serves as a significant statistical mediator between learners’ positive attitudes toward technology and their perceived effectiveness of AI-supported English language learning. The research, conducted by Amr M. Mohamed and Fawaz Al Mahmud, examined 234 students enrolled in English-language courses at a private college of health sciences in Saudi Arabia. The study specifically investigated whether self-efficacy, defined as the belief in one’s ability to manage prospective situations, explains the relationship between technophilia and positive emotional experience on one hand, and perceived learning effectiveness on the other. The findings indicate that confidence in managing AI-supported tasks is a central cognitive mechanism linking affective dispositions to learners’ appraisals of educational utility.

The study employed a quantitative, cross-sectional correlational design to examine associations among the constructs without manipulating variables. This approach allowed the researchers to analyze the interrelationships and hypothesized mediation pathways within a single integrated framework. The theoretical model drew on the Technology Acceptance Model, Control-Value Theory, and Social Cognitive Theory. These theories collectively suggest that stable technological dispositions and dynamic emotional experiences influence perceived learning effectiveness through proximal cognitive beliefs, specifically self-efficacy. The integration of these frameworks allows for a comprehensive view of how learners interact with AI tools in educational settings.

Participants were recruited through convenience sampling from intact classes at a private health sciences college with a nursing focus. The sample consisted of 234 students, with a mean age of 20.85 years and a standard deviation of 4.06. The gender distribution was 77.8% male and 22.2% female, a demographic imbalance the authors noted may limit the generalizability of the findings to female-majority nursing populations. Academic levels included 128 undergraduates and 106 foundation-year students. Regarding AI tool use for English learning, 41.5% of participants reported daily use, 29.9% weekly, 10.3% monthly, 13.2% rarely, and 5.1% no use, indicating varied levels of engagement with the technology.

Data were collected using a structured self-administered questionnaire composed of five sections, with the majority using 5-point Likert-type scales. Technophilia was measured using four items adapted from Martínez-Córcoles et al., demonstrating a Cronbach’s alpha of 0.87. Positive emotional experience was assessed with four items adapted from Bieleke et al., yielding a Cronbach’s alpha of 0.91. Self-efficacy was evaluated using three context-specific items based on Bandura’s theoretical framework, with a Cronbach’s alpha of 0.82. Perceived learning effectiveness was measured using a scale adapted from Ashfaq et al.; after removing one item with a negative factor loading, the final three-item scale demonstrated a Cronbach’s alpha of 0.927. Instrument validation included exploratory and confirmatory factor analyses, which supported the unidimensional structure of the scales.

Structural equation modeling with bootstrapping, using 5,000 resamples, was employed to examine direct and indirect associations. The model fit indices indicated acceptable fit, with a chi-square to degrees of freedom ratio of 1.75, a Comparative Fit Index of 0.941, and a Root Mean Square Error of Approximation of 0.057. The structural model explained 52% of the variance in perceived learning effectiveness and 48% of the variance in self-efficacy. Preliminary analyses confirmed that measurement invariance was supported across gender groups, ensuring that the constructs were measured comparably among male and female participants. Multicollinearity diagnostics indicated that tolerance and variance inflation factor values were within acceptable ranges, suggesting that multicollinearity did not substantially affect the estimation of structural relationships.

The results showed that self-efficacy demonstrated the strongest direct association with perceived learning effectiveness, with a standardized path coefficient of 0.375 and a p-value of 0.001. Positive emotional experience was also strongly associated with self-efficacy (β = 0.624, p < 0.001), and its indirect association with perceived learning effectiveness through self-efficacy was significant (β = 0.234, 95% CI [0.128, 0.351]). Technophilia was positively associated with self-efficacy (β = 0.307, p = 0.002), and its indirect association with perceived learning effectiveness through self-efficacy was also statistically significant (β = 0.115, p = 0.005). These findings support the hypothesis that self-efficacy partially mediates the relationships between both technophilia and positive emotional experience with perceived learning effectiveness.

Direct associations also indicated that technophilia (β = 0.270, p = 0.001) and positive emotional experience (β = 0.281, p = 0.023) were significantly associated with perceived learning effectiveness independently of self-efficacy. The persistence of these direct effects alongside significant indirect effects confirms the pattern of partial statistical mediation. This suggests that while confidence in using AI is a crucial factor, positive orientations toward technology and enjoyable emotional experiences also contribute directly to learners’ perceptions of effectiveness. The partial mediation implies that other psychological mechanisms, such as perceived cognitive load or trust in AI-generated feedback, may also play a role in shaping these evaluations.

Control variable analyses revealed that frequency of AI tool use showed significant positive associations with both perceived learning effectiveness (β = 0.146, p = 0.012) and self-efficacy (β = 0.132, p = 0.016). This indicates that more frequent use of AI tools was associated with greater confidence and more favorable appraisals of learning effectiveness. In contrast, gender was not significantly associated with perceived learning effectiveness (β = 0.082, p = 0.221) or self-efficacy (β = 0.064, p = 0.294), suggesting that the relationships examined in the model did not differ systematically according to gender in this specific sample. The lack of significant gender differences supports the robustness of the mediation model across the gender groups present in the data.

The authors interpret these findings through the lens of Social Cognitive Theory, which posits that self-efficacy regulates effort allocation, persistence, and strategic action. In AI-mediated learning environments, perceived ability is decisive when learners must interpret automated responses, select appropriate tools, and organize learning activities with agency. Learners with high self-efficacy are more likely to apply feedback strategically and sustain performance under higher task demands. The study suggests that self-efficacy functions as a proximal cognitive mechanism through which AI affordances become usable learning opportunities, rather than merely a motivational belief. This perspective highlights the importance of confidence in managing the complexities of AI-supported language learning.

Several limitations were acknowledged by the researchers, primarily the cross-sectional design, which prevents causal inference and restricts the interpretation of the relationships to associational rather than directional. The reliance on self-report measures increases the likelihood of common-method bias, although statistical checks using the common latent factor approach suggested this was not a significant issue. The study focused on perceived learning effectiveness, a subjective evaluation, rather than objective performance outcomes, which limits the generalizability of the findings to actual language proficiency gains. Additionally, the single-institution convenience sampling and the gender imbalance in the nursing student population further constrain the applicability of the results to broader EFL contexts.

The study contributes to computer-assisted language learning research by clarifying how positive orientation toward AI-supported technologies, emotional experience, and efficacy beliefs jointly relate to learners’ perceptions of effective AI-supported language learning. The findings suggest that pedagogical strategies should focus on progressively scaffolded practice and mastery-oriented AI tasks to strengthen self-efficacy. By building learners’ confidence in managing AI-supported tasks, educators may improve perceived effectiveness appraisals in AI-supported EFL environments. The research underscores that while exposure to AI tools is important, the psychological mechanisms linking technological dispositions to learning outcomes are critical for effective educational integration.

Future research should prioritize experimental and longitudinal designs that manipulate or track self-efficacy and examine its influence on objective AI-supported language-learning outcomes. Qualitative or mixed-methods research could also investigate how learners develop, maintain, or lose confidence over time during authentic interactions with different AI tools. These approaches would clarify the temporal and potentially causal processes underlying the relationships identified in this study. The current findings provide a foundational understanding of the psychological dynamics at play in AI-supported EFL education, highlighting the pivotal role of self-efficacy in shaping learners’ experiences and evaluations.

Subject of Research: Psychological mechanisms linking technophilia and emotional experience to perceived learning effectiveness in AI-supported EFL education

Article Title: The mediating role of self-efficacy in the relationships of technophilia and positive emotional experience with perceived learning effectiveness in AI-supported EFL education

Article References: Mohamed, A. M., & Al Mahmud, F. (2026). The mediating role of self-efficacy in the relationships of technophilia and positive emotional experience with perceived learning effectiveness in AI-supported EFL education. Discover Artificial Intelligence, 6(1), Article 1737. https://doi.org/10.1007/s44163-026-02712-z

Image Credits: AI Generated

DOI: 10.1007/s44163-026-02712-z

Keywords: AI-supported EFL education, self-efficacy mediation, technophilia, positive emotional experience, perceived learning effectiveness, structural equation modeling, nursing students, cross-sectional survey, Technology Acceptance Model, Control-Value Theory, Social Cognitive Theory, computer-assisted language learning, psychological dispositions, AI tool usage frequency, learner confidence