The integration of AI in medical and dental education has garnered considerable attention, with growing evidence supporting its role in enhancing student engagement, comprehension, and self-directed learning [9, 10, 36]. The present study reinforces these perspectives, demonstrating that AI-assisted terminology support significantly improved comprehension among Linguistically diverse dental students in oral anatomy. The statistically significant improvement from 10.4 to 16.1 in comprehension scores, coupled with enhanced engagement levels, supports the premise that AI can serve as an effective supplement to traditional pedagogy, particularly for students facing linguistic barriers.

From a theoretical standpoint, the findings are well-aligned with both Self-Determination Theory (SDT) and Cognitive Load Theory (CLT) [11, 13]. CLT posits that instructional design should reduce unnecessary mental effort (extraneous load) so that learners can devote cognitive resources to meaningful processing [11]. In this study, AI reduced linguistic barriers, enabling students to concentrate on core anatomical concepts rather than struggling with unfamiliar terminology. The positive impact on student engagement also aligns with SDT’s dimensions of autonomy and competence, as students experienced greater control over their learning and confidence in navigating challenging content [13]. For example, the predominance of term clarification queries (45% of total) directly illustrates learners’ attempts to reduce extraneous cognitive load as posited by CLT [11]. The significant gains in absorption and dedication scores map onto SDT’s constructs of competence and autonomy [13], reflecting genuine motivational benefits.

Crucially, these results address a notable research gap. While prior studies have focused on AI’s use in diagnostic simulation and image recognition in dental education, few have explored its capacity to support terminology learning for non-native English-speaking students [9, 36]. The current findings extend this literature by demonstrating that AI can scaffold terminology acquisition in context-sensitive and adaptive ways—particularly valuable in EMI-based and multilingual education settings. This also aligns with Airey’s (2011) concept of disciplinary literacy, which emphasises the need for students not only to acquire technical vocabulary but also to develop fluency in the discourse practices specific to their academic discipline. By simplifying and contextualising complex anatomical terms, ChatGPT may support learners in navigating the linguistic demands of disciplinary knowledge in EMI environments [7].

This study also confirms that student engagement is not merely a by-product of novelty but reflects authentic motivation, confidence, and clarity in learning. The increase in UWES-S engagement scores—corroborated by qualitative insights—suggests that AI-supported learning, when appropriately structured, can promote deeper learner investment and classroom participation.

Thematic findings from the focus group interviews closely aligned with quantitative results, demonstrating effective triangulation [22]. For example, students who reported increased comprehension and confidence in survey responses also shared during focus groups that ChatGPT helped them “check understanding immediately” and “learn at [their] own pace.” These insights contextualized the upward shifts in UWES-S engagement scores and terminology test performance. From a theoretical perspective, the reduction in linguistic and cognitive strain supports Cognitive Load Theory (CLT) by suggesting that AI scaffolding reduced extraneous load, allowing students to focus on core anatomical content [11, 17]. This is consistent with recent work suggesting that AI can support deeper conceptual learning by alleviating language and cognitive barriers [18]. Furthermore, the emphasis on self-paced exploration and improved confidence aligns with Self-Determination Theory (SDT), particularly the satisfaction of autonomy and competence needs [13], echoing prior research on AI-mediated environments that foster motivation and learner agency [19]. These findings may also be informed by learner personality traits. As Özbey and Yasa (2025) argue, individual variation in personality significantly affects how students perceive and engage with AI in educational settings [14]. Students with high openness may be more willing to experiment with ChatGPT, while more anxious or less extroverted students may exhibit caution or resistance. This aligns with the broader literature on personality in dental education, where traits such as Conscientiousness and Agreeableness are associated with stronger academic performance and learner autonomy [15]. Furthermore, Li et al. (2024) found that dominant MBTI profiles among dental students—particularly ISTJ and ESTJ types—often prefer structured, reliable information sources, which may explain varied patterns of interaction with ChatGPT’s open-ended, iterative interface [16].

Nonetheless, ethical considerations merit emphasis. As students themselves noted, AI-generated content may vary in accuracy, and unchecked reliance could foster passive learning or misinformation [19, 20]. Faculty involvement thus remains essential to guide appropriate AI use, verify information accuracy, and model critical appraisal [20, 23]. Without such oversight, students may adopt AI as a shortcut rather than a scaffold.

Furthermore, the study surfaces important implications regarding institutional preparedness. The need for faculty training in AI literacy is crucial to prevent digital divides and ensure equitable learning. This aligns with emerging calls for health professional education institutions to co-design AI integration strategies that balance innovation with pedagogical responsibility [20, 28].

This pilot study demonstrates that AI tools like ChatGPT can provide accessible, personalized support for linguistically diverse learners, particularly in subjects with dense terminology such as oral anatomy. However, it also underscores that AI should not be viewed as a standalone solution. A hybrid AI-human approach, incorporating faculty feedback, ethical policies, and curriculum alignment, is essential to ensure responsible and sustainable AI adoption [29].

Finally, while the results are promising, they are tempered by limitations, including a small sample size, the absence of a control group, and reliance on self-reported data. These were acknowledged and mitigated through methodological triangulation [22], but they nevertheless restrict generalizability. Additionally, we acknowledge that factors such as students’ prior English language proficiency, digital literacy levels, and pre-existing familiarity with AI platforms may have influenced both engagement and comprehension outcomes. These potential confounding variables were not systematically assessed in this pilot study and should be addressed in future research designs to enhance validity and generalizability. Future research should investigate the longitudinal effects of AI-supported learning across diverse educational settings and examine the perspectives of educators, whose roles are pivotal in shaping the AI-learning ecosystem [37]. They should also explore how learner personality traits shape interaction patterns, trust, and cognitive engagement with AI tools like ChatGPT, building on the findings of recent scholarly works [14,15,16]. Studies should systematically measure and adjust for potential confounders such as prior English proficiency and digital literacy to strengthen validity and interpretability of findings. While purposive sampling enabled the inclusion of students most relevant to the study aims, this strategy limits representativeness. Future research will aim for more diverse sampling strategies, including random or stratified approaches, to enhance generalizability [24, 35].

Conclusion and implications for practice

This study provides empirical evidence that AI-assisted terminology learning, particularly through ChatGPT, can enhance comprehension and engagement among linguistically diverse dental students. AI-enabled tools align well with both Cognitive Load Theory and Self-Determination Theory by reducing linguistic cognitive load and supporting learner autonomy. These technologies offer promising scaffolds for self-directed learning and can be thoughtfully integrated into multilingual, EMI-based programs where terminology mastery is essential.

Importantly, this study highlights that AI integration must be pedagogically sound and ethically guided. While AI can support learner independence and access to knowledge, unchecked use may lead to superficial understanding or overreliance on potentially inaccurate outputs [18, 19]. Faculty involvement and structured institutional policies remain critical to model responsible AI use, foster critical thinking, and mitigate risks.

For educators and institutions, this research underscores the value of AI as a supplemental—not substitutive—tool within guided learning ecosystems. As AI platforms become increasingly accessible, structured training initiatives must be implemented to build AI literacy among both students and faculty. Attention to digital equity, algorithmic transparency, and ethical safeguards will be essential to ensure inclusive, context-sensitive adoption.

Future research should explore the longitudinal impacts of AI-supported learning across diverse cultural and linguistic contexts, the role of faculty in co-mediating AI engagement, and the development of adaptive AI models aligned with disciplinary depth and learner diversity [36, 37]. It should also explore faculty perspectives and institutional readiness for AI integration, as co-designed approaches are critical to developing sustainable, context-sensitive educational strategies. As this field matures, dental and health professions education must lead by example in ensuring that AI technologies are deployed not only for innovation’s sake, but for the advancement of equity, understanding, and human learning.