An artificial intelligence application framework for nursing education: a scoping review based on the KAS model Authors: Yue Xiang, Haoning Shi, Keke Ren, Xilin Yang, Lin Li, Qinghua Zhao, Huanhuan Huang Published: 2026 ABSTRACT BackgroundArtificial intelligence (AI) is reshaping education, but existing AI application frameworks cannot be effectively applied to nursing education, which is defined by its focus on practical skills and clinical judgment.AimsThis study aims to address this gap by synthesizing literature on current educational AI application frameworks to propose a dedicated AI application framework for nursing education, grounded in the core nursing competency model of Knowledge, Attitude, and Skills (KAS) to guide curriculum design.MethodsA systematic search was conducted in PubMed, Embase, Web of Science, and CNKI to identify literature on educational AI application frameworks. The search strategy combined controlled vocabulary and free-text keywords, adapted to the indexing structure of each database. Reference lists of eligible studies and relevant articles were manually screened to supplement the electronic search. The search was updated to March 2026.ResultsThe study systematically screened the initial 1750 retrieved documents, ultimately including 23 studies for analysis. The included publications spanned the years 2022 to 2026 and covered 15 countries worldwide. Through in-depth analysis of these documents, the study identified and synthesized an AI application framework for nursing education comprising three major dimensions: knowledge, attitudes, and skills. This framework encompasses 12 key sub-dimensions, providing detailed guidance for curriculum design in nursing education, and is proposed for an AI application framework tailored to nursing education.ConclusionThe potential of AI in nursing education is transformative. To fully unlock this potential and ensure responsible application, existing research should focus on empirically validating the proposed framework through empirical studies, exploring adaptive implementation strategies for resource-constrained settings, and developing educators’ AI application capabilities.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12909-026-09209-4. Introduction In today’s rapidly evolving digital landscape, AI has emerged as a transformative force, reshaping various aspects of modern lives [1, 2]. AI is now profoundly reshaping education systems worldwide [3–6]. Currently, AI’s integration into education, manifested through intelligent tutoring systems, adaptive language-learning chatbots, and workflow optimizations, has enhanced data processing and task efficiency, positioning AI as an indispensable educational companion [7–9]. However, despite the increasing application of AI in education, existing AI application frameworks cannot effectively guide curriculum design for the highly specialized field of nursing education. Recent bibliometric analysis reveals a sustained upward trend in AI research within nursing education [10]. As a critical domain for cultivating healthcare professionals, nursing education emphasizes practical application, rigorous clinical skill development, and complex ethical decision-making [11, 12]. Therefore, specializing current educational AI application frameworks to meet the unique needs of nursing education has become a critical yet underexplored research direction. Theoretical foundations The KAS model originated from mid-20th-century educational psychology research and was developed by scholars such as Bloom, building on the Taxonomy of Educational Objectives [13, 14]. This model categorises learning outcomes into three interrelated core dimensions: knowledge acquisition within the cognitive domain (Knowledge, K), attitude formation within the affective domain (Attitude, A), and skill mastery within the psychomotor domain (Skill, S). Following decades of refinement and development, the KAS model has become a widely applied theoretical framework in education, particularly in professional competence development [7, 15]. In nursing education, the KAS model is particularly significant [16]. The nursing profession demands not only that learners acquire a solid foundation in theoretical knowledge, but also that they cultivate an appropriate professional attitude and mastery of clinical skills [17, 18]. Extensive research indicates that outstanding nursing professionals must attain high standards across all three dimensions of knowledge, attitude and skill [19–21]. Any deficiency in these areas may compromise the quality of nursing care and patient safety [22, 23]. Numerous studies have confirmed that AI application in education can similarly be mapped onto the three dimensions of the KAS model. This empirical evidence encompasses nursing education and is also widely distributed across general education domains [24, 25]. Research aims and questions The aim of this study is to employ a scoping review to systematically map and critically analyse AI application frameworks in global education, using the KAS model as a lens to synthesize the literature. The review addresses the following research questions: What are the key components and thematic emphases of current educational AI application frameworks?How can these components be mapped and analysed through the lens of the Knowledge, Attitude, and Skill (KAS) model?Based on the analysis of existing frameworks using the KAS model, what core principles and curriculum design guidelines can be derived to cultivate AI competency among nursing students? What are the key components and thematic emphases of current educational AI application frameworks? What are the key components and thematic emphases of current educational AI application frameworks? How can these components be mapped and analysed through the lens of the Knowledge, Attitude, and Skill (KAS) model? How can these components be mapped and analysed through the lens of the Knowledge, Attitude, and Skill (KAS) model? Based on the analysis of existing frameworks using the KAS model, what core principles and curriculum design guidelines can be derived to cultivate AI competency among nursing students? Based on the analysis of existing frameworks using the KAS model, what core principles and curriculum design guidelines can be derived to cultivate AI competency among nursing students? Methodology Research design This study employed the JBI scoping review methodology [26]. The protocol was registered with PROSPERO (CRD420251118417). To ensure methodological rigour and transparency, this review was conducted in strict adherence to the JBI Manual for Evidence Synthesis and reported according to the PRISMA-ScR checklist [27], as shown in Appendix S1. Literature search strategy From June 2025 to August 2025, we systematically searched four databases: PubMed, Embase, Web of Science, and CNKI. An update of this initial search was performed in March 2026. The search strategy was developed through discussions within the research group and was subsequently tailored to the specific indexing requirements of each database. It was built around key terms such as “Education,” “Artificial Intelligence,” “AI,” “Framework,” “Guideline,” “Application,” and “Practice,” which were combined using Boolean operators and adapted to each database’s specific syntax. Based on preliminary PubMed search results, we developed a search strategy and adjusted it according to the characteristics of each database. Concurrently, we manually screened and included relevant research references. The search strategy and results for all databases are shown in Appendix S2. Literature screening The literature screening process strictly adhered to the PCC framework (Table 1) recommended by the JBI Manual for scoping reviews to formulate the inclusion criteria. The Covidence systematic review software platform was utilized for duplication and initial screening. Subsequently, two researchers independently performed the title/abstract screening and the full-text screening. Any disagreements were resolved through discussion or by consulting a third researcher. Table 1Population, concept, and contextPCC ElementsDefinitionPopulation (P)Refers to the target populations, including educators, and learners.Concept (C)Refers to the research concept, focusing on literature that describes educational AI application frameworks, models, or implementation evidence.Context (C)Refers to the setting of the study. Given the identified gap in nursing AI application frameworks, the context was strategically broadened to include general education, medical education, and other relevant professional disciplines. Table 1Population, concept, and contextPCC ElementsDefinitionPopulation (P)Refers to the target populations, including educators, and learners.Concept (C)Refers to the research concept, focusing on literature that describes educational AI application frameworks, models, or implementation evidence.Context (C)Refers to the setting of the study. Given the identified gap in nursing AI application frameworks, the context was strategically broadened to include general education, medical education, and other relevant professional disciplines. Population, concept, and context Inclusion criteria The target population encompasses educators and learners; the concept entails an explicit description of an AI application framework in education, outlining its objectives, core features, and implementation examples; the context encompasses general education, medical education, and other relevant professional disciplines; the required language for description is either Chinese or English. Exclusion criteria Full-text unavailable, duplicates, incomplete data, or studies lacking a clearly articulated educational AI framework. Data analysis and extraction Data were extracted independently by two reviewers using a co-developed form capturing bibliographic details and framework elements; disagreements were resolved by consensus or a third reviewer. The analysis employed a deductive-inductive hybrid approach based on the KAS model, refining the three dimensions of knowledge, skill, and attitude into twelve sub-dimensions to guide the design of nursing AI curricula. The coding process involved preliminary coding, thematic grouping, and refinement, conducted independently by two researchers with iterative cross-checking to ensure consistency. Methodological transparency is ensured through the mapping and extraction matrix provided in Appendix S3. Results Literature screening process A total of 1750 articles were retrieved from the 4 databases, as illustrated in the PRISMA flow diagram (Fig. 1). After removing 322 duplicates, 6 of the included studies have been retracted, 1197 articles were excluded based on title and abstract screening. A total of 225 full-text articles were assessed for eligibility. Ultimately, 23 studies met the inclusion criteria and were selected in this review. Fig. 1PRISMA flow diagram PRISMA flow diagram Basic characteristics for inclusion in the literature As shown in Table 2, the 23 studies included in this review spanned 15 countries globally, with the majority originating from the United States (n = 5), China (n = 3), the United Kingdom (n = 2), and Turkey (n = 2); the remaining 11 countries each contributed one study. In terms of publication years, the studies spanned from 2022 to 2026, with the highest proportion published in 2025 (n = 14), which clearly indicates that research in this field is experiencing sustained and rapid growth over time. Regarding research subjects, studies focusing on learners were most prevalent (n = 12), followed by those centered on educators (n = 5), while 6 studies addressed both groups. In terms of research design, the included studies comprised 15 reviews, 4 qualitative studies, and 4 quantitative studies. Analyzing application domains, current AI application frameworks discussions are most concentrated in medical education (n = 11), followed by general education (n = 9), and three studies that focused on other professional disciplines such as computer science and environmental design. Table 2Characteristics of Included Studies (N = 23)Characteristics n %Publication year 202628.7% 20251460.9% 2024417.4% 202328.7% 202214.3%Country United States521.7% China313.0% United Kingdom28.7% Turkey28.7% Malaysia14.3% Colombia14.3% Ireland14.3% Japan14.3% New Zealand14.3% Kosovo14.3% Sweden14.3% Italy14.3% Austria14.3% Iran14.3% Australia14.3%Target population Learners1252.2% Educators521.7% Both626.1%Study design Review1565.2% Qualitative417.4% Quantitative417.4%Application domain Medical education1147.8% General education939.1% Other education313.0% Table 2Characteristics of Included Studies (N = 23)Characteristics n %Publication year 202628.7% 20251460.9% 2024417.4% 202328.7% 202214.3%Country United States521.7% China313.0% United Kingdom28.7% Turkey28.7% Malaysia14.3% Colombia14.3% Ireland14.3% Japan14.3% New Zealand14.3% Kosovo14.3% Sweden14.3% Italy14.3% Austria14.3% Iran14.3% Australia14.3%Target population Learners1252.2% Educators521.7% Both626.1%Study design Review1565.2% Qualitative417.4% Quantitative417.4%Application domain Medical education1147.8% General education939.1% Other education313.0% Characteristics of Included Studies (N = 23) Analysis of the KAS dimensions in educational AI application frameworks Theoretical framework for AI applications in current education Among the 23 included studies, the present study identifies an evolutionary trend in educational AI application frameworks shifting from a focus on singular technological cognition to comprehensive competency development and deep integration with ethical and contextual dimensions. Although existing frameworks differ in objectives and emphases, their core components generally encompass the fundamental dimensions of Knowledge, Attitude, and Skill. However, these elements remain scattered across various literature sources, lacking systematic synthesis and comparison. To address this, this study first constructs a KAS model comprising three primary dimensions and twelve consistently emerging secondary sub-dimensions (Table 3). Building upon this framework, existing models are then decomposed to clearly illustrate their distribution and emphasis across each KAS dimension (Fig. 2; Table 3). Table 3Analysis and Distribution of KAS Dimensions in Educational AI Application FrameworksPrimary DimensionSecondary Sub-dimensionDescriptionReferencesKnowledge (K)Foundational AI Knowledge (K1)Understanding basic AI concepts, terminology, and core threshold concepts.[1, 28–42]Technical AI Knowledge (K2)Understanding the technical components, working mechanisms, and algorithmic foundations of AI.[1, 29, 31– 34, 36, 42, 43]AI Ethics Knowledge (K3)Understanding ethical issues (e.g., fairness, accountability, transparency, privacy) and societal impacts in AI applications.[1, 30, 31, 33, 34, 36, 38, 39, 41, 43–47]AI Application Knowledge (K4)Understanding application scenarios, potential value, and associated risks/limitations of AI in specific domains.[1, 28, 29, 33, 34, 44, 45, 47, 48]Attitude (A)Critical Thinking (A1)Maintaining a cautious approach to AI outputs, actively analyzing their strengths, limitations, biases, and inaccuracies.[1, 30, 31, 33, 35, 36, 40, 44–46, 48]Responsibility (A2)Taking active responsibility for the entire process and consequences of AI use, focusing on data privacy, error identification, and contingency handling.[1, 30, 31, 34, 38–45, 47, 49]Ethical Values (A3)Valuing and practicing ethical principles such as fairness, transparency, inclusivity, and accountability.[1, 29, 31, 32, 37, 38, 42–44, 47, 49]Openness & Adaptability (A4)Actively accepting and adapting to AI technology, and willingly participating in its integration and innovation.[1, 29, 30, 36–39, 41, 43, 44, 46]Skills (S)Operational Skills (S1)Ability to use AI tools and systems, mastering basic skills like prompt engineering and interface operation.[31, 37–44]Application Skills (S2)Ability to apply AI tools to solve practical problems in areas such as teaching and research.[1, 29–31, 33–35, 37, 38, 40, 42, 43, 45–48]Collaborative Skills (S3)Ability to engage in effective human-AI collaboration and interdisciplinary cooperation.[28, 30–33, 43, 49]Evaluation Skills (S4)Ability to assess the quality, accuracy, and risks of AI outputs and make professional judgments.[1, 29–35, 38, 39, 41–48] Table 3Analysis and Distribution of KAS Dimensions in Educational AI Application FrameworksPrimary DimensionSecondary Sub-dimensionDescriptionReferencesKnowledge (K)Foundational AI Knowledge (K1)Understanding basic AI concepts, terminology, and core threshold concepts.[1, 28–42]Technical AI Knowledge (K2)Understanding the technical components, working mechanisms, and algorithmic foundations of AI.[1, 29, 31– 34, 36, 42, 43]AI Ethics Knowledge (K3)Understanding ethical issues (e.g., fairness, accountability, transparency, privacy) and societal impacts in AI applications.[1, 30, 31, 33, 34, 36, 38, 39, 41, 43–47]AI Application Knowledge (K4)Understanding application scenarios, potential value, and associated risks/limitations of AI in specific domains.[1, 28, 29, 33, 34, 44, 45, 47, 48]Attitude (A)Critical Thinking (A1)Maintaining a cautious approach to AI outputs, actively analyzing their strengths, limitations, biases, and inaccuracies.[1, 30, 31, 33, 35, 36, 40, 44–46, 48]Responsibility (A2)Taking active responsibility for the entire process and consequences of AI use, focusing on data privacy, error identification, and contingency handling.[1, 30, 31, 34, 38–45, 47, 49]Ethical Values (A3)Valuing and practicing ethical principles such as fairness, transparency, inclusivity, and accountability.[1, 29, 31, 32, 37, 38, 42–44, 47, 49]Openness & Adaptability (A4)Actively accepting and adapting to AI technology, and willingly participating in its integration and innovation.[1, 29, 30, 36–39, 41, 43, 44, 46]Skills (S)Operational Skills (S1)Ability to use AI tools and systems, mastering basic skills like prompt engineering and interface operation.[31, 37–44]Application Skills (S2)Ability to apply AI tools to solve practical problems in areas such as teaching and research.[1, 29–31, 33–35, 37, 38, 40, 42, 43, 45–48]Collaborative Skills (S3)Ability to engage in effective human-AI collaboration and interdisciplinary cooperation.[28, 30–33, 43, 49]Evaluation Skills (S4)Ability to assess the quality, accuracy, and risks of AI outputs and make professional judgments.[1, 29–35, 38, 39, 41–48] Analysis and Distribution of KAS Dimensions in Educational AI Application Frameworks Fig. 2Distribution of discussion frequency across dimensions in the included literature Distribution of discussion frequency across dimensions in the included literature Knowledge dimensions K1 emphasizes the understanding of fundamental concepts and terminology, which is recognized as the foundation by most research [1, 28–42]. For example, Allen et al. [28] specifically highlighted the necessity of exploring the “concept of thresholds”. Mousavi et al. [41] used Braun and Clarke’s thematic analysis to establish ‘Definitions, History, and Types of AI’ as core educational foundations. K2 requires an understanding of the technical components and working mechanisms, as detailed in 9 publications [1, 29, 31–34, 36, 42, 43] that support the K2 requirement, such as the underlying principles of generative AI and similar technologies. K3 is the most widely recognized dimension within the K-dimensional framework, with 14 studies [1, 30, 31, 33, 34, 36, 38, 39, 41, 43–47] discussing this dimension, focusing on core ethical issues such as fairness, accountability, transparency, and privacy. Lan et al. [33] established Technological Ethical Knowledge (GenAI-TEAK) as a strong predictor of AI integration (β = 0.636, p < 0.001), empirically confirming ethics knowledge as a critical driver of effective teaching rather than just a theoretical constraint. The assessment scenarios and potential risks addressed by K4 are examined in 9 studies [1, 28, 29, 33, 34, 44, 45, 47, 48] to establish awareness of technological boundaries and societal impacts. Attitude dimensions A1 was examined in 11 studies [1, 30, 31, 33, 35, 36, 40, 44–46, 48]. Notably, Greengrass [30] proposed a specific method to cultivate this skill by analyzing erroneous AI outputs. A2 and A3 each received 14 papers [1, 30, 31, 34, 38–45, 47, 49] and 11 papers [1, 29, 31, 32, 37, 38, 42–44, 47, 49]. A2 holds particular significance in medical education, embodying proactive responsibility for the entire AI application process and its consequences. An et al. [49] is grounded in a systematic review of 17 empirical studies. This dimension emphasizes the user’s lens where practitioners exercise subjective agency to take active responsibility for identifying and mitigating risks such as bias and data privacy throughout the AI interaction process. Ayci et al. [37] utilizing structural equation modeling, demonstrated that a positive Perceptual Future Orientation significantly enhances students’ Ethical Values (β = 0.308, p < 0.05). This is pivotal for safeguarding patient safety and privacy. A4 is a prerequisite for embracing technology, with its necessity highlighted in 11 studies [1, 29, 30, 36–39, 41, 43, 44, 46], which underscore its indispensability. For instance, Qehaja [43] argues that as AI evolves, maintaining agile adaptability is crucial to seizing emerging opportunities and addressing challenges. Skill dimensions Despite its significance, S1 is explicitly addressed in only 9 publications [31, 37–44], where it is concretized as a new skill set requiring mastery, such as ‘prompt engineering’. S2 and S3 received 16 papers each [1, 29–31, 33–35, 37, 38, 40, 42, 43, 45–48] and 7 papers [28, 30–33, 43, 49]. Bautista et al. [38] provides empirical granularity to S2 by identifying seven specific generative AI use cases in nursing practice where drafting care plans and discharge instructions solve practical problems. Specifically, Kobeissi et al. [32] prioritizes interdisciplinary collaboration, advocating that educators co-develop curricula with experts in computer science, law, and ethics to facilitate effective AI understanding and safe implementation in nursing education. S4 represents the skill garnering the broadest consensus, with 18 studies [1, 29–35, 38, 39, 41–48] addressing this skill. Among these, Erkayiran [39] provided critical empirical evidence within nursing education. Their quasi-experimental study demonstrated that AI could evaluate therapeutic communication skills with high reliability (ICC > 0.97). The example from [34] utilized Retrieval-Augmented Generation (RAG) to automate the evaluative function, cross-referencing data to detect cognitive biases, while explicitly reserving the final decision for the teacher’s professional judgment. Application of AI application frameworks in nursing education The 23 studies selected for analysis represent a diverse educational landscape, strategically incorporating general, interdisciplinary, and medical contexts to construct a comprehensive framework. To ensure theoretical robustness, we included 9 studies from general education and 3 from other disciplines (e.g., Computer Science, Environmental Education). These were essential for providing theoretical scaffolding and unique technical insights that are often absent in purely clinical literature. These cross-disciplinary perspectives enrich the framework’s adaptability and depth. Notably, Medical Education emerged as the dominant application domain. Within this cluster, Nursing Education demonstrated a distinct prominence, accounting for 6 studies. This high proportion highlights nursing education’s proactive responsiveness to technological paradigm shifts, signaling that nursing education is currently spearheading the practical integration of AI within the broader healthcare education sector. Discussion This study employs a scoping review methodology to systematically organize and summarize current AI application frameworks in education based on the KAS model for the first time. The findings suggest that although AI is extensively applied in education, there is a pronounced absence of AI application frameworks tailored specifically to nursing education. From a temporal perspective, publications from 2025 accounted for the highest proportion (60.9%) among the literature included in this study. This reflects that research into educational AI application frameworks is undergoing rapid development, representing an emerging and increasingly prominent field of inquiry. This trend underscores the urgent need for nursing education to keep pace with these advancements by actively exploring the integration of AI technologies, thereby guiding the cultivation of nursing professionals equipped for the evolving healthcare landscape. However, this skewed temporal distribution also indicates that research conclusions in this domain require further validation and refinement through long-term studies. Regarding geographical distribution, existing research exhibits a pronounced regional concentration, with the United States (21.7%) and China (13%) accounting for a significant proportion of research outputs. This distribution pattern reveals the reality of the digital divide, where the development of educational AI application frameworks is closely linked to regional economic and technological levels. Notably, Chinese scholars explicitly advocate for localized AI application frameworks through a pedagogy-oriented approach to ensure cultural and resource feasibility [45]. Extending this insight to nursing education, targeted strategies are needed to mitigate such disparities. Firstly, nursing educators may adopt teaching strategies that do not rely on continuous access to advanced AI systems. For example, offline approaches such as case-based discussions and conceptual demonstrations of AI workflows, combined with locally adapted or simplified AI models, particularly those supporting local languages, can support learners from diverse contexts, thereby promoting equitable participation and reducing the risk of marginalization. Secondly, this study recommends that nursing curriculum design should prioritize KAS sub-dimensions that are less dependent on advanced technical infrastructure. Within the K dimensions, emphasis should be placed on K1 and K3, rather than technical K2, enabling students to grasp AI’s implications without requiring sophisticated equipment. Regarding the A dimensions, fostering A1, A2, and A3 empowers nurses to maintain clinical autonomy and professional judgment, an ability that is essential for patient safety when using tools trained on non-representative datasets. Within the S dimensions, the focus should shift toward S4 and S3, rather than complex S1, while S2 application skills should be taught with flexibility according to technological access. By emphasizing these adapted priorities, and centering professional nursing judgment in evaluating AI outputs, this adapted framework contributes to inclusive AI competency development and helps prevent the marginalization of low-resource regions during the technological transition, aligning with nursing’s core values of equity, compassion, and evidence-based practice. By mapping the KAS elements across the reviewed literature, we found that some dimensions (specifically K1,K3,A2,S2,S4) attracted wide consensus, whereas others were scarcely addressed. This uneven distribution reveals the current focus and developmental trajectory of educational AI application frameworks. For instance, K1 and K3 garnered extensive attention, a characteristic aligning with the emphasis on safe practice foundations within nursing practice. Nurses are expected to possess a solid grasp of AI foundational knowledge and AI ethical risks knowledge (e.g., privacy, bias) as a prerequisite for exercising independent professional judgement. Equally noteworthy is that S2 and S4 account for a substantial share of current educational AI application frameworks. In nursing education, S2 is essential for the practical application of AI to clinical tasks (e.g., care planning)and S4 provides the critical evaluation needed to detect algorithmic hallucinations. Furthermore, A2 garnered substantial support, resonating with the core ethical principles of nursing practice. As nurses bear ultimate responsibility for patient safety and care quality, they must maintain subjective agency even when employing AI-assisted decision-making. Therefore, AI application in nursing education must the cultivation of professional responsibility to ensure practitioners remain the final authority in clinical care. This aligns precisely with the professional accountability emphasized in the International Council of Nurses (ICN) Code of Ethics [50]. In contrast to the highly discussed dimensions, technical and operational competencies (K2, K4, S1) were explicitly addressed in only 9 publications. The possible reasons are threefold: these dimensions involve a high degree of technical professionalism that is inherently difficult for students and educators to understand and master; the pronounced regional disparities in AI infrastructure render such competencies less accessible in many low-resource settings; and more importantly, most existing frameworks have tended to prioritize foundational, ethical, and evaluative dimensions over technical ones, a focus that supports the curriculum design proposed in this study. However, nurses need to master technical competencies (K2, K4, S1) to ensure clinical safety, as literature [51] explicitly states that digital skills training for healthcare workers must include concrete operational competencies to ensure technologies are applied safely and effectively. To address these gaps, future educational strategies could incorporate simulation-based learning approaches, such as virtual reality (VR) or high-fidelity clinical simulations, to provide experiential training in AI-supported clinical operations [52]. Importantly, in resource-constrained settings, low-cost alternatives may provide feasible entry points for developing operational competencies, including screen-based simulations with pre-recorded cases, open-access AI platforms (e.g., online model demonstration tools), and basic AI literacy activities such as chatbot-based learning and guided exercises using publicly available datasets. Furthermore, S3 received support from only 7 studies. This is likely because current research prioritizes discrete functional competencies such as S2 and S4, which focus on individual task execution and output validation, while human-AI and interdisciplinary collaboration models remain underdeveloped and not yet fully operationalized within curricula. Nevertheless, numerous studies [53, 54] indicate that nurses must participate in shaping the future of human-machine collaboration and engage in active interdisciplinary collaboration. This role presupposes core competencies for synergistic work with AI systems and provides a theoretical rationale for nursing-education models that advance “nurse–AI–health-care-team” collaboration. To operationalize S3, nursing curricula may integrate interdisciplinary, team-based learning activities (e.g., AI-assisted case discussions and role-play), alongside collaborative projects and joint problem-solving tasks in simulated clinical contexts. It is worth noting that existing research has begun exploring frameworks for AI applications in nursing education. For instance, the ‘N.U.R.S.E.S.’ framework proposed by Hoelscher and Pugh [31], which emphasizes nurses’ roles as AI collaborators, and the ‘AI-ABCs’ framework by Kobeissi et al. [32] clarifies fundamental educational objectives. Additionally, the ‘CARE’ framework recently introduced by Bester and Zeigler [40] structures prompt engineering proficiency through context, action, role, and expectation elements. However, these frameworks still face challenges in translating into concrete curricula. They either lack clear pathways for pedagogical implementation or focus on isolated skills rather than comprehensive instructional content and assessment methodologies, thereby creating difficulties for educators seeking to translate them into actionable and holistic curricula. As visually summarized in Fig. 3, the proposed AI application framework translates the KAS model into a framework with twelve actionable sub-dimensions. Unlike isolated models, this framework assimilates the theoretical breadth of general education and the technical diversity of specialized fields. Specifically, it incorporates threshold concepts from Computer Science and virtual scenario applications from Environmental and Architectural education while remaining deeply rooted in the clinical context of medical education.Under comprehensive background, the AI application framework for nursing education developed based on the KAS model demonstrates significant advantages. This framework not only aligns with contemporary demands by responding to global ethical calls regarding AI and the urgent need for AI application in nursing practice but more importantly, it provides a clear solution. By refining the three dimensions of knowledge, attitudes, and skills into 12 specific sub-dimensions, it offers educators clear guidance for systematically designing teaching objectives, organizing instructional content, and conducting teaching evaluations. The establishment of this framework holds profound significance for cultivating nursing professionals capable of adapting to intelligent healthcare environments, as highlighted in the International Council of Nurses’ 2025 report on the 78th World Health Assembly, which urges Member States to “harness safe and effective digital health and artificial intelligence opportunities” and emphasizes the urgent need for nurses to acquire the skills and competencies required to deliver high-quality, person-centred care in a digital health environment [55]. By integrating elements such as critical thinking, ethical decision-making, and evaluation skills into the curriculum, it ensures nursing students not only operate AI tools but also make ethically sound professional judgement within complex clinical contexts. Fig. 3Proposed KAS Framework for AI Integration in Nursing Education Proposed KAS Framework for AI Integration in Nursing Education To fully unlock the potential of AI in nursing education and ensure its responsible application, future research should focus on three key directions. First, empirical studies must be conducted to validate the framework’s effectiveness in enhancing nursing students’ AI application skills across diverse clinical contexts. Second, Future research must explore adaptive implementation strategies for resource-constrained settings to ensure the digital divide does not widen during the global transition to AI-integrated nursing. Finally, significant emphasis must be placed on cultivating the AI application capabilities of educators to ensure they can effectively utilize the framework for teaching. This study has several limitations. Firstly, the literature review was confined to Chinese and English databases, potentially overlooking significant research from other linguistic and cultural contexts, thereby affecting the comprehensiveness of the findings. Secondly, the rapid advancement of artificial intelligence technologies, coupled with the continuous emergence of new applications and ethical considerations, necessitates ongoing updates to the framework to maintain its timeliness. Conclusions This scoping review systematically analysed the current status and developmental trends of AI application frameworks within the educational domain. We categorized the included literature and conducted an in-depth analysis of the core elements of existing frameworks across three dimensions: knowledge, attitude, and skill. Findings indicate that despite significant advancements in educational AI application frameworks in recent years, a notable gap persists between these frameworks and the practical demands of nursing education. Existing frameworks lack operational guidance for curriculum translation, thereby identifying a crucial direction for future research. Furthermore, this study identifies substantial regional disparities in the maturity of AI educational frameworks. This imbalance underscores the need to consider regional adaptability alongside the promotion of universal frameworks. The AI application framework for nursing education, proposed based on the KAS model, provides specific curriculum design guidelines for cultivating nursing professionals with comprehensive AI competency. This holds significant importance for advancing the deep integration of AI with nursing education, safeguarding patient safety, while explicitly reaffirming the nurse’s role as the final authority in clinical decision-making in AI-assisted contexts. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3.