We adopt the framework by Arksey and O’Malley, which includes 5 key stages in the implementation of a scoping review [12]. This framework was selected to map the existing literature, identify research gaps, and provide an overview of the available evidence related to the research objectives. The methodological structure used consists of 5 stages: stage 1—defining the research question; stage 2—identifying studies relevant to the research question; stage 3—selecting studies to be included in the review; stage 4—mapping data from the included studies; and stage 5—synthesizing, summarizing, and reporting the findings.

The research question for this scoping review is: how has ChatGPT been used as an educational resource in nursing practice, and what educational outcomes and implementation challenges have been reported?

A systematic search was conducted across 3 electronic databases: PubMed, Scopus, and ProQuest. The search was performed on (January 19, 2025) to identify relevant literature published in the last 6 years.

The inclusion and exclusion criteria for this review were established according to the Population, Concept, Context model. The search strategy was built using the Population, Concept, Context framework with specific keywords and Boolean operators as lists (Textbox 1).

The search was limited to studies published between 2019 and 2025, written in English, and with full-text availability.

Data from the included studies were systematically extracted into a standardized data-charting form developed by the team. The following variables were extracted: (1) authors and publication year, (2) country of origin, (3) study design, (4) study population and sample size, (5) specific AI technology or intervention investigation (ChatGPT), (6) form of educational resource, and (7) key findings relevant to the review objectives. The data charting form was piloted on 5 included studies and refined accordingly [6,8,14-16]. The article synthesis process was supported by an AI-based tool (Elicit.ai), which was used to assist with organizing and summarizing study findings, while the final interpretation and thematic grouping were conducted by the researchers. Data synthesis was conducted using a descriptive and iterative thematic grouping approach rather than formal qualitative coding, in accordance with the exploratory purpose of scoping reviews

The synthesized data are presented in Table 1.

The findings from the literature were used to identify outcomes based on emerging keywords. This review will analyze all articles through titles, abstracts, and full texts, followed by a check to identify any differences and duplicates. All articles analyzed contain information regarding AI technology as a source of nursing education.

This study was exempted from the evaluation requirements by the Institutional Review Board for Human Subjects because the articles analyzed only disclosed anonymized information or did not contain personal data that could identify specific individuals.

This scoping review included 20 studies [6,8,14-31] published between 2019 and 2025, conducted across diverse geographic regions (Table 2). The predominance of studies from Asia reflects varying levels of AI adoption and research maturity rather than differences in effectiveness across health care systems. A detailed breakdown of the geographic distribution of the included studies is provided (Table 2).

Overall, 20 articles were analyzed, and various research designs were used to evaluate the application of AI in health care. Findings were interpreted narratively by study design to provide a clearer understanding of the strength and nature of the available evidence (Table 2).

Across the included studies, AI was applied in nursing practice and education primarily as a supportive educational and informational tool, rather than as an autonomous clinical decision-maker [17,19]. The AI technologies identified can be broadly categorized into three functional groups: (1) large language model–based systems, (2) machine learning (ML)–driven analytic tools, and (3) interactive educational technologies [8,16,27].

First, large language models, particularly ChatGPT (GPT-3.5 and GPT-4), were the most frequently reported AI tools [6,19,29,30]. These systems were used to provide on-demand explanations, support information-seeking behaviors, and assist with educational clarification for nurses, patients, and caregivers [24,25,32]. Studies consistently described ChatGPT as facilitating access to health information, supporting learning activities, and assisting communication in both educational and clinical contexts[6,24,29].

In parallel, several studies reported the use of ML, natural language processing (NLP), and robotic process automation to support clinical monitoring, risk prediction, and information management [8,17,20]. These applications focused on enhancing early detection, optimizing workflow efficiency, and supporting decision-making processes by synthesizing large volumes of clinical data [21,26]. Importantly, these technologies were described as augmenting, rather than replacing, professional judgment [17,21].

A third group of applications involved interactive educational technologies, including AI-based chatbots, virtual avatars, educational robots, AI-generated e-books, and virtual or augmented reality tools [16,22,31]. These technologies were primarily used in nursing education to support collaborative learning, simulation-based training, and communication skill development, particularly in pediatric and remote learning contexts [16,22,31].

Overall, the reviewed studies indicate that AI technologies are implemented heterogeneously across nursing contexts, with their role largely confined to information support, educational facilitation, and workflow assistance [7,19,32]. Direct clinical decision-making autonomy was not reported, and professional oversight remained a consistent requirement across applications [17,21,25].

Across the included studies, reported outcomes related to patient education varied considerably in scope and methodological rigor. Educational outcomes associated with AI-supported patient education were reported across cognitive, behavioral, and informational domains, although the depth and rigor of outcome measurement varied substantially between studies [7,19]. Consequently, outcome evidence was interpreted cautiously and weighted according to study design (Table 3).

Quantitative evidence of cognitive outcomes was primarily reported in quasi-experimental studies. A controlled study in Taiwan demonstrated that integrating ChatGPT into nursing education significantly improved critical thinking (mean 4.73, SD 0.44), problem-solving skills (mean 4.53, SD 0.71), and learning satisfaction (mean 4.69, SD 0.46) compared with a control group (P<.001) [6]. These findings suggest that large language models can enhance higher-order learning processes when embedded within structured educational designs. Trend and bibliometric analyses further indicated a rapid increase in the application of AI-based educational tools, including chatbots and educational robots, within nursing and Science, Technology, Engineering, and Mathematics education since 2018 [16]. While these analyses do not measure learning outcomes directly, they reflect growing institutional and academic interest in AI-supported education.

Several studies evaluated informational outcomes by assessing improvements in the readability and comprehensibility of patient education materials. ChatGPT was shown to reduce the reading level of medical texts from grade 11 to grade 9 while largely preserving content integrity, whereas Google Bard achieved lower reading levels at the expense of substantial content reduction [30]. These findings highlight the potential of AI to improve accessibility of patient education materials, while also underscoring variability across AI tools. Similarly, an AI-assisted educational intervention for ostomy patients demonstrated high understandability (mean 81.91%, SD 19.05%) and actionability (mean 85.33%, SD 26.44%) scores using the Patient Education Materials Assessment Tool (Agency for Healthcare Research and Quality) [28]. These outcomes suggest that AI may support tailored patient education in contexts requiring structured and repetitive informational support.

In clinical and caregiving settings, AI-supported education was associated with improved perceived quality, clarity, and usefulness of health information. Comparative evaluations of large language model–based tools found that ChatGPT outperformed alternative platforms in accuracy, completeness, and clinical use when responding to caregiver information needs, particularly in pediatric oncology contexts [24]. However, these studies emphasized that AI-generated information should be interpreted as supportive, rather than definitive, educational guidance. Clinical education–related outcomes were reported in disease-specific contexts. In Alzheimer disease management, GPT-4 generated responses rated by neurologists as significantly higher in quality than those produced by GPT-3.5 (P<.001), suggesting potential value in supporting caregiver education for complex chronic conditions [29]. Nonetheless, these findings were based on controlled evaluations and did not assess downstream clinical or behavioral outcomes.

Behavioral and psychosocial outcomes were primarily reported in qualitative and development studies. AI-generated educational e-books significantly reduced fear-related behavioral responses among pediatric patients undergoing medical procedures (β=−1.540; P <.05), while also enhancing nursing students’ self-efficacy in therapeutic communication [22]. In teleconsultation settings, HCPs reported improved engagement and efficiency in patient education, although concerns regarding diagnostic accuracy and contextual appropriateness persisted [25]. Overall, patient education outcomes associated with AI use were most consistently demonstrated in controlled educational settings and condition-specific interventions. However, the predominance of descriptive and short-term studies limits conclusions regarding sustained clinical impact. Collectively, the evidence suggests that AI-supported patient education may contribute to improved psychosocial and behavioral outcomes in specific contexts, although evidence of sustained impact remains limited.

This scoping review highlights various implementations of AI technology in nursing education and clinical practice, reviewing 20 studies that assess the impact of AI in multiple aspects of nursing. The key findings from this review can be categorized into four main areas; (1) AI in patient education and simplification of medical information [17,23,24,28,30,31], (2) AI in clinical decision-making and patient monitoring [8,20,29], (3) AI in nursing education [6,15,16,18,19,22,27] and (4) challenges and prospects of AI in nursing [14,21,25,26](Figure 2).

To enhance clarity and readability, the main findings of this review are summarized in Table 4, highlighting 4 key areas of AI application in nursing along with the associated evidence.

AI has been used to simplify medical information to make it more understandable for patients. Studies show that ChatGPT and Google Bard can adjust the language level in medical materials to make them more accessible to patients from diverse educational backgrounds [30]. Furthermore, AI has been used in developing patient education materials, as found in research on ostomy patients, where AI-generated materials were easier to understand and helped patients better comprehend their care [28]. This indicates that AI can play a crucial role in improving health literacy and assisting patients in making better decisions about their care.

Generative AI technology and NLP-based chatbots are increasingly used in nursing education to enhance nursing students’ critical thinking, problem-solving, and communication skills. Studies show that ChatGPT can help students better understand nursing concepts and increase their engagement and enjoyment of learning [6].

Additionally, AI robots and chatbots have supported game-based learning and distance learning, which are becoming increasingly popular in medical and nursing education. Research trends show a rise in the use of AI in nursing education since 2018, with a primary focus on technology-based learning [19]. Meanwhile, AI-based virtual patients have also improved nursing students’ self-efficacy and confidence in communicating with patients. Nursing students practicing with AI systems demonstrated enhanced communication skills, which are crucial to their future clinical practice [22].

AI technologies based on ML, NLP, and robotic process automation have been instrumental in early disease detection, clinical decision-making, and patient monitoring. Studies show that AI can enhance diagnostic accuracy and expedite the decision-making process for nurses and doctors [8]. In neurology, GPT-4 in managing Alzheimer disease also demonstrates that AI can generate accurate medical information and assist in diagnosis and disease management. A comparison between AI responses and those from neurologists indicates that AI has the potential to become a valuable medical support tool in managing neurodegenerative diseases [29].

Although AI holds significant potential in patient education within nursing care settings, several challenges must be addressed. Ethical issues, data privacy, information accuracy, and the limitations of AI in providing personalized approaches to patient education are significant concerns in its implementation. Some studies indicate that while AI can simplify medical information and enhance patient understanding, there are concerns that AI may diminish the humanistic aspect of communication between health care providers and patients [26]. Patient education relies not only on the information provided but also on emotional support and empathy, aspects that AI-based systems may not fully replace. Furthermore, another challenge is how AI can ensure the accuracy of the information provided, as large language model-based AI systems, such as ChatGPT and Google Bard, may sometimes generate invalid details or information that is less appropriate for clinical contexts [30].

Improve information accessibility, cognitive engagement, and perceived educational quality across diverse nursing contexts. However, outcome measurement remains inconsistent, with relatively few studies employing robust experimental designs or long-term follow-up. Most reported outcomes reflect short-term or context-specific benefits, highlighting the need for further research to evaluate sustained educational impact, behavioral change, and integration into routine nursing practice.

In teleconsultation, AI has helped improve communication efficiency between health care providers and patients, particularly in explaining medical conditions and treatment procedures. However, challenges remain in data security and legal responsibility, particularly regarding how the information provided by AI is managed and to what extent AI can be relied upon to deliver accurate medical education [25]. Additionally, in long-term care, research shows that AI is more widely accepted when used to monitor patient health rather than replace direct communication between health care providers and patients. This suggests that AI can enhance the efficiency of patient education systems. However, it must still be integrated with human interaction to ensure that patients fully understand medical information and feel emotionally supported during care [21].

This scoping review demonstrates that current applications of ChatGPT and related AI technologies in nursing are predominantly supportive in nature, functioning as adjunct educational tools rather than autonomous systems within clinical practice. Overall, the findings indicate that AI is predominantly positioned as a supportive educational and informational tool, rather than as an autonomous clinical decision-maker. This role alignment is consistent across diverse nursing contexts and reflects the early and exploratory stage of AI integration in health care education. Therefore, this review highlights 3 key aspects of AI application in nursing education: interpretation of AI use, the challenges encountered, and the implications of AI in nursing practice.

Across diverse contexts, AI was primarily used to facilitate information access, educational clarification, and learner engagement, particularly under conditions of limited time and workforce capacity [7,25]. Rather than replacing professional judgment, AI tools are commonly described as supportive resources that improve efficiency and responsiveness, particularly in clinical and educational settings with limited time and high workload demands [20,33]. This pattern aligns with implementation science principles, which emphasize that digital innovations are most effective when embedded within existing professional workflows and governed by human oversight [17,18].

Similarly, ML-based systems and analytical tools have been applied to assist clinical monitoring, risk assessment, and decision support activities in nursing practice [21,27]. Importantly, the literature consistently emphasizes that these technologies are intended to support, rather than replace, nursing expertise. This reinforces the ongoing importance of professional responsibility and clinical judgment in patient education and care delivery, even as AI tools become more widely adopted [8,15].

Despite promising findings, the evidence base remains constrained by heterogeneity in study design and outcome measurement. Many studies relied on descriptive or qualitative approaches, limiting causal inference and generalizability. One of the main barriers is the lack of clear regulations and guidelines regarding the use of AI in health care services [20,34]. This lack of precise regulation leads to variations in the effectiveness of AI implementation across different health care and educational institutions. Some studies also indicate that AI has the potential to produce inaccurate or even misleading information, particularly in the context of telemedicine and AI-based consultations [25]. This presents a risk to patients who rely on AI as their primary source of medical information.

Additionally, ethical issues and data privacy are significant challenges in using AI in nursing education. This study reveals that the use of AI-based chatbots in mental health services continues to face challenges related to patient data security and the potential misuse of medical information provided by AI [35]. Therefore, measures are needed to ensure that the AI technology used in nursing education meets high standards of security and accuracy. From an implementation perspective, the findings suggest that AI adoption in nursing education should be accompanied by structured training, institutional guidelines, and continuous evaluation to ensure safe and effective integration into practice [26].

Overall, the findings of this scoping review suggest that AI-supported education has the potential to support nursing education and patient engagement, particularly as a complementary educational resource alongside nurse-led instruction. However, the available evidence remains limited and heterogeneous. Therefore, future research should prioritize implementation-focused studies, longitudinal designs, and standardized evaluation frameworks to assess the sustained educational and clinical impact of AI technologies in nursing practice.

Based on the results of this scoping review, the following recommendations are proposed to guide future research, policy development, and implementation efforts in nursing education and practice.

This scoping review has several limitations that should be considered when interpreting the findings. While challenges related to AI implementation, such as ethical concerns, data privacy, and the potential generation of inaccurate information, are frequently discussed in the included literature, these issues were not empirically evaluated within this review and therefore could not be assessed for their direct impact on educational effectiveness.

In addition, the findings of this scoping review are influenced by limitations in the search strategy and inclusion criteria. The literature search was restricted to studies published between 2019 and 2025, written in English, and available in full text, which may have resulted in the exclusion of relevant studies published in other languages or earlier periods.

Furthermore, only 3 electronic databases were searched, which may have limited the comprehensiveness of the evidence captured. Although this approach aligns with scoping review methodology, it may have led to the omission of relevant gray literature or discipline-specific publications.

The included studies also demonstrated substantial heterogeneity in study design, context, AI technologies, and outcome measures, which limited the ability to compare findings across studies or draw conclusions regarding effectiveness. As a result, the synthesis was descriptive rather than evaluative, consistent with the exploratory purpose of a scoping review.

The scoping review compares its findings with prior work, noting that while AI, particularly ChatGPT, shows significant promise in nursing education, the technology faces challenges similar to those outlined in earlier studies. Previous research also points to concerns about AI’s capacity to replace human interaction in patient education. The review emphasizes that AI’s effectiveness in improving clinical decision-making and patient monitoring has been widely acknowledged, but further development and regulation are necessary to address the ethical and accuracy concerns highlighted by earlier work.

This scoping review mapped the emerging evidence on the use of ChatGPT as an educational resource in nursing practice and education. The findings indicate that ChatGPT is primarily applied as a supportive, adjunct tool to facilitate information access, educational clarification, and learner engagement, rather than as an autonomous or decision-making system.

However, the current evidence base is heterogeneous and largely exploratory, with limited empirical evaluation of long-term educational or clinical outcomes. Ethical considerations, data privacy, and the need for professional oversight remain critical challenges for implementation.

Future research should prioritize implementation-oriented and longitudinal studies to clarify how ChatGPT and similar AI technologies can be responsibly integrated into nursing education and practice. In particular, standardized outcome measures, clear governance frameworks, and evaluation of professional oversight mechanisms are needed to assess sustained educational value and contextual appropriateness across diverse nursing settings.