Transforming Nursing Education with Artificial Intelligence: A Systematic Review (2010–2025)

Abstract

Introduction

This systematic review provides the first comprehensive synthesis of empirical studies on Artificial Intelligence (AI) integration in nursing education, offering actionable insights for nurse educators and clinical leaders. It highlights how AI transforms learning environments by enhancing personalization, feedback, and instructional efficiency.

Aims

To examine how AI is applied across nursing education settings and its impact on learning outcomes.

Methods

A systematic search of PubMed, CINAHL, IEEE Xplore, and Scopus identified peer-reviewed studies published from January 2010 to April 2025. Eligible studies focused on empirical AI applications in academic, clinical, or hybrid nursing education contexts. Studies were appraised using the Critical Appraisal Skills Programme (CASP) checklist, and findings were synthesized thematically.

Results

Twenty-eight studies met the inclusion criteria. AI-enhanced nursing education in four main areas: (a) personalized learning systems tailored content to individual needs, (b) simulation-based training improved decision-making in high-acuity scenarios,(c) automated assessment tools provided immediate, unbiased feedback, and (d) at the institutional level, AI supported curriculum management and predictive analytics. Common risks included technological inequities, faculty preparedness gaps, and ethical concerns around privacy and bias.

Conclusion

To support implementation, this study recommends: (a) integrating AI-powered simulation into emergency care training, (b) deploying adaptive platforms to support at-risk learners, and (c) using automated tools for real-time formative feedback. Diagnostic accuracy is proposed as a measurable outcome to assess impact. The next step for educators is to initiate multi-site pilot programs over 6–12 months, evaluating improvements in learning outcomes, trust, and system integration.

Introduction

Nursing education serves as the backbone of healthcare systems, equipping future nurses with the clinical, cognitive, and ethical competencies necessary for safe and effective patient care (Ma et al., 2025; Zhou et al., 2024). Traditional pedagogical methods, rooted in classroom instruction, clinical practicums, and simulation labs, are increasingly challenged by the growing complexity of patient needs, workforce shortages, and the accelerating pace of healthcare innovation (Lifshits & Rosenberg, 2024). In response to these pressures, nursing education requires adaptive and scalable strategies that enhance learning efficiency, clinical preparedness, and instructional effectiveness (Seibert et al., 2021). AI has emerged as a transformative tool to address these evolving educational demands by bridging gaps in supervision, feedback, and personalization (Buchanan et al., 2021; Hwang et al., 2024).

AI technologies encompass a wide range of applications within nursing education. These include intelligent tutoring systems for personalized learning, machine learning platforms for performance prediction, virtual and augmented reality tools for clinical simulation, and automated systems for assessment and feedback (Abdelwahab et al., 2025; Harmon et al., 2021; Jallad et al., 2024). AI modalities such as natural language processing, generative models like ChatGPT, decision-support tools, and rule-based algorithms are increasingly being deployed across educational contexts to assist both learners and educators (De Gagne et al., 2024; Nesa et al., 2025; Shorey et al., 2019). The scope of this review includes populations such as pre-licensure nursing students, nurse educators, and continuing education participants across academic, clinical, and hybrid learning environments (Montejo et al., 2024; Topaz et al., 2025; Tran et al., 2024).

Although interest in AI integration has surged in healthcare, much of the existing research has concentrated on clinical applications such as diagnostics and robotic surgery, rather than the pedagogical use of AI in nursing education (Luo et al., 2024; O’Connor et al., 2023). Previous reviews often aggregate findings from broader medical education contexts, making it difficult to isolate outcomes that are specific to nursing and its unique professional competencies (Schneidereith & Thibault, 2023; Von Gerich et al., 2022). Additionally, there has been insufficient attention to how AI impacts essential educational dimensions like knowledge retention, diagnostic reasoning, learner equity, and faculty workload (Ahmed, 2024; Jung, 2023; Lebo & Brown, 2024). This gap in consolidated evidence limits the ability of educators, administrators, and policymakers to make informed decisions about AI integration that align with nursing pedagogy and ethics (Colborn et al., 2023; Rasouli et al., 2021).

The present review sought to fill this gap by providing a structured synthesis of current empirical evidence on the integration of AI in nursing education (Gunawan et al., 2024; Martinez-Ortigosa et al., 2023). It evaluates the educational utility of AI tools across various settings and modalities and organizes findings through three practical lenses: their impact on learning outcomes, their role in improving instructional workflow, and their ethical and safety considerations, including data privacy, algorithmic fairness, and equity in access (Castonguay et al., 2023; Foronda & Porter, 2024). By mapping emerging patterns and persistent challenges, the review aims to offer a balanced, evidence-informed perspective that guides the responsible adoption of AI in nursing curricula (Farsi, 2025). It emphasizes the importance of aligning technology with pedagogical goals, preparing faculty for implementation, and safeguarding humanistic values such as empathy, communication, and critical thinking. In doing so, this work contributes to a future-ready vision for nursing education in the digital age.

Methodology

Study Design

This systematic review was conducted to thoroughly examine how AI is being integrated into nursing education. The systematic review methodology was chosen because of its rigorous approach and its ability to compile and analyze data from a wide range of study designs and sources. This ensures a broad and in-depth understanding of existing literature. By providing a structured and unbiased evaluation of published research, this method is essential for assessing the various ways AI technologies are applied in nursing education and their overall impact. The review design is particularly justified given the diversity and novelty of AI applications in educational contexts, which require comprehensive aggregation and thematic analysis across qualitative, quantitative, and mixed-methods research. Outcomes of interest were defined a priori and included knowledge acquisition, skills development, learner engagement, diagnostic reasoning, faculty workload, instructional efficiency, and educational equity. This review explored both the scope and depth of AI integration in nursing education, identified gaps in existing research, and provided recommendations for future studies. To provide a clear representation of the study selection process, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram was used to illustrate the number of records identified, screened, excluded, and included in the final synthesis, as shown in Figure 1 (Page et al., 2021). This systematic selection process ensured that only the most relevant and credible studies contributed to the findings of this review.

Figure 1.

PRISMA diagram.

Search Strategy

The literature search was conducted across multiple electronic databases to ensure a comprehensive collection of relevant studies. These databases included PubMed, CINAHL, IEEE Xplore, and Scopus, all well-regarded for their extensive coverage of medical and technological research (Table 1). To capture all pertinent studies, carefully selected search terms were used, incorporating combinations of keywords such as “artificial intelligence,” “nursing education,” “medical education,” “AI in healthcare education,” “AI tools in education,” “technology in education,” and “technological innovation in nursing.” To maintain consistency in language interpretation, the search was limited to studies published in English. Additionally, the publication timeframe was restricted from January 2010 through April 2025, allowing the review to focus on the most recent advancements in the field. Grey literature was excluded to ensure the inclusion of peer-reviewed studies only, thereby maintaining a higher level of methodological rigor and reliability in the synthesis.

Table 1.

Search Strategy.

Database	Search String	Filters	Last Search Date	Screening Roles	Disagreement Resolution	Inclusion Criteria	Exclusion Criteria
PubMed	(“artificial intelligence” OR “AI”) AND (“nursing education” OR “medical education” OR “AI in healthcare education” OR “AI tools in education” OR “technology in education” OR “technological innovation in nursing”)	English; 2010–2025; Peer-reviewed only	March 25	Two independent reviewers: Title/abstract and full-text screening	Disagreements resolved by consensus; third reviewer consulted if unresolved	Empirical studies on AI in nursing education; Studies involving nursing students or educators; AI applications with measurable outcomes; Peer-reviewed; English language	Non-empirical articles; Editorials/opinions; Non-English; Not related to educational outcomes
CINAHL	(“artificial intelligence” OR “AI”) AND (“nursing education” OR “clinical education” OR “healthcare training” OR “e-learning” OR “simulation training” OR “virtual learning”)	English; 2010–2025; Peer-reviewed only	March 25	Two independent reviewers: Title/abstract and full-text screening	Disagreements resolved by consensus; third reviewer consulted if unresolved	Focus on AI-driven tools in nursing curricula; Original research (quantitative/qualitative/mixed-methods); Published in English; Peer-reviewed	Grey literature; Non-education studies; Studies without nursing-specific data; Reviews without primary data
Scopus	(“artificial intelligence” OR “AI”) AND (“nursing education” OR “digital learning” OR “intelligent tutoring” OR “adaptive learning” OR “clinical simulation” OR “virtual patients”)	English; 2010–2025; Peer-reviewed only	April 25	Two independent reviewers: Title/abstract and full-text screening	Disagreements resolved by consensus; third reviewer consulted if unresolved	Studies on AI use in academic/clinical settings with nursing participants; English; Peer-reviewed	Studies focused solely on general medical education without extractable nursing data; Opinion pieces; Conference abstracts
IEEE Xplore	(“artificial intelligence” OR “machine learning”) AND (“nursing education” OR “AI in training” OR “educational AI systems” OR “automated assessment” OR “virtual simulation”)	English; 2010–2025; Peer-reviewed only	April 25	Two independent reviewers: Title/abstract and full-text screening	Disagreements resolved by consensus; third reviewer consulted if unresolved	Studies examining AI technologies used in educational platforms for nurses; Peer-reviewed articles in English	Technical/engineering papers without educational context; Non-peer-reviewed; Non-English

Inclusion and Exclusion Criteria

The inclusion criteria were carefully designed to focus on original studies that explore the application of AI in educational curricula. This included empirical research offering insights into the effectiveness, challenges, and outcomes of AI implementation in educational settings. Eligible populations were nursing students, nursing educators, or both, in academic, clinical, or online learning contexts. Studies were included if they examined AI interventions such as intelligent tutoring systems, virtual simulations, adaptive learning platforms, or AI-driven assessment tools, with outcomes related to knowledge, skills, competencies, attitudes, or educational performance. Both quantitative and qualitative study designs were considered, provided they were peer-reviewed and met minimum quality standards based on CASP assessment. Interprofessional or medical education studies were included only if data specific to nursing could be clearly identified. To maintain a high standard of evidence, exclusion criteria were established to omit studies not directly related to educational improvement, as well as opinion pieces and editorials. This approach ensured that only data-driven, peer-reviewed research was considered for analysis.

Selection and Screening Process

A structured approach was employed to identify and include the most relevant studies. The initial phase involved screening titles and abstracts to evaluate their alignment with the research questions. Subsequently, a comprehensive full-text review was conducted to determine eligibility based on predefined inclusion and exclusion criteria. To enhance efficiency and organization, EndNote was utilized for reference management and to streamline the screening process. To ensure rigor and minimize bias, two reviewers independently screened all titles/abstracts and full texts against the eligibility criteria. Disagreements were resolved through discussion, and when consensus could not be reached, a third reviewer was consulted. Inter-rater agreement during the screening process was assessed using Cohen's κ statistic, which demonstrated substantial agreement (κ = 0.82).

Data Extraction

Data were systematically extracted from each selected study, emphasizing key elements such as authors, publication year, study objectives, sample size, methods, key findings, recommendations, and types of AI utilized (Table 2). This structured approach ensured a comprehensive and organized evaluation of the collected data. To synthesize the extracted information, findings were tabulated, and a thematic analysis was conducted to identify common themes, emerging trends, and notable outliers within the data.

Table 2.

Features of the Selected Studies.

Authors and Publication Year	Country	Study Objectives	Sample Size	Methods	Key Findings	Recommendations	Types of AI Utilized
Wood et al. (2021)	USA	To assess medical student and faculty attitudes toward AI, in preparation for teaching AI foundations and data science applications in clinical practice in an integrated medical education curriculum.	121 medical students, 52 clinical faculty	Online 15-question semi-structured survey	Only 30% of students and 50% of faculty were aware of AI in medicine. Majority learned about AI from media. Faculty reported lower AI literacy than students. Interest in AI education is high among both groups.	Medical schools should integrate AI into curricula. AI literacy should be enhanced through interdisciplinary teaching teams. More research is needed to structure AI education effectively.	Survey focused on AI literacy and AI applications in medical education (not specific AI tools).
Sridharan and Sequeira (2024)	Kingdom of Bahrain	To explore the application of AI tools in classroom instruction and student assessment using a pharmacology & therapeutics case study.	Not applicable (AI-generated content assessment)	Descriptive proof-of-concept cross-sectional study using generative AI tools (ChatGPT, Sage Poe, Claude-Instant)	AI tools generated structured learning objectives, test items (MCQs, SAQs, OSPEs), and assessment blueprints. AI-generated outputs showed homology but required expert validation for content accuracy.	AI tools can assist in medical education by optimizing instructional methods and assessment strategies, but expert validation is essential before implementation.	ChatGPT 3.5, Sage Poe (Anthropic), Claude-Instant (Anthropic)
Jackson et al. (2024)	India	To evaluate medical students’ perceptions of AI in medicine, their preferences for AI training in education, and their grasp of AI's ethical implications in healthcare.	325 medical students	Cross-sectional survey using a pre-validated, semi-structured questionnaire.	72.3% of students agreed that AI reduces medical errors, and 54.2% believed AI enhances medical decision accuracy. 37.6% feared AI could replace physicians, while 69.2% were concerned about its impact on the humanistic aspect of medicine. 91.4% had not received formal AI training.	Medical education should integrate structured AI training. Emphasis should be placed on reducing medical errors and addressing ethical concerns. Further curriculum development is needed to improve AI literacy among medical students.	No specific AI tools were utilized; study focused on perceptions of AI applications in medical education.
Sami et al. (2025)	Pakistan	To assess students’ perceptions regarding the credibility and effectiveness of AI as a learning tool and explore AI integration in medical education.	702 medical students	Cross-sectional study using a 26-question survey, analyzed with SPSS v26.0.	80.3% of students had a favorable attitude toward AI in education. 60.8% considered AI an effective learning tool, while 58.4% found it credible. 66.8% reported AI provided more accurate answers compared to traditional study tools. Students viewed traditional tools as becoming increasingly outdated (59%).	Develop dedicated AI tools for medical education. Integrate AI formally into medical curricula with regulatory oversight. Ensure AI enhances human learning rather than replacing traditional methods.	ChatGPT, MetaAI, Google Gemini
Kidwai et al. (2025)	Netherlands	To assess pharmaceutical science students’ perspectives on the integration of machine learning (ML) training in undergraduate curricula.	15 students	Semi-structured interviews conducted over three years with students who participated in the ML module.	Students valued the module as well-designed and effective. 80% had minimal to no prior ML knowledge. Many students expressed a strong desire to continue ML training in their careers. Students highlighted the importance of integrating ML into pharmaceutical education.	Integrate ML training into pharmaceutical sciences curricula to prepare future drug researchers. Provide more guidance at the start of ML modules. Break down complex ML concepts into smaller segments to enhance learning.	Machine Learning (ML) applications in biomarker discovery and predictive analytics.
Rony et al. (2024a)	Bangladesh	To investigate healthcare workers’ concerns about artificial intelligence replacing medical professionals and its implications for job security, ethics, and healthcare policy.	33 healthcare professionals from diverse settings	Descriptive and exploratory qualitative research using the Technology Acceptance Model, Technology Threat Avoidance Theory, and Sociotechnical Systems Theory. Data collected via semi-structured interviews and focus group discussions.	Seven key themes emerged: job security concerns, trust in AI, ethical dilemmas, impact on patient care, role redefinition, patient-provider relationships, and policy regulation. Healthcare workers expressed anxiety about job displacement but acknowledged AI's potential to enhance decision-making. Ethical concerns centered on maintaining human compassion in healthcare. Policy frameworks are essential to balance AI integration with quality healthcare delivery.	Develop robust AI policies ensuring job security and ethical AI use. Enhance AI literacy training for healthcare professionals. Ensure AI complements rather than replaces human expertise. Strengthen legal frameworks to define AI responsibilities and liabilities.	AI applications in clinical decision-making, diagnostics, patient management, and predictive analytics.
Issa et al. (2024)	Jordan, UAE, KSA, Egypt	To examine AI-related knowledge, attitudes, and perceived challenges of healthcare profession education (HPE) students regarding AI integration.	642 medical, nursing, physiotherapy, and clinical nutrition students from four public universities	Cross-sectional study using an online survey covering AI benefits, impact on trust, impact on healthcare professionals, AI inclusion in HPE, and AI integration challenges.	66.4% of participants reported low AI knowledge, with the UAE having the highest proportion (72.7%). 54.4% learned about AI outside their curriculum, mainly from social media (66%). 51.2% expressed positive attitudes towards AI integration, with Egypt showing the highest support (59.1%). 91% viewed AI positively in healthcare, but 43.5% believed AI could negatively impact patient trust.	Enhance AI awareness and structured training in HPE curricula. Encourage interdisciplinary collaboration for AI education. Address ethical and policy concerns to ensure responsible AI adoption in healthcare education.	Survey-based study; examined AI applications in healthcare education but did not implement specific AI tools.
Blease et al. (2022)	Ireland	To assess the experiences and opinions of final-year medical students regarding their exposure to AI/ML during their degree.	252 final-year medical students from four medical schools in Ireland	Cross-sectional paper-based survey administered after compulsory final-year classes.	66.5% of students reported zero hours of AI/ML teaching in their degree. 43.4% had not heard of the term ‘machine learning’. 80.6% had not read any academic journal articles on AI/ML. 41.1% planned to learn about AI/ML, and 78.6% agreed that AI/ML discussion should be part of medical training.	Medical schools should develop short, cross-disciplinary courses in AI/ML. Training should include critical thinking about AI, ethics, and biases. Medical ethics courses should integrate AI-related discussions, and AI literacy should be improved among students.	Study focused on AI/ML exposure and education; no specific AI tools were implemented.
Salem et al. (2024)	Saudi Arabia	To examine the relationship between nursing students’ personality traits and their attitudes toward AI.	218 nursing students from three governmental universities in Saudi Arabia	Multicenter cross-sectional study using an online survey with the Big Five Inventory and the General Attitudes toward AI Scale.	Nursing students with high openness scores had positive attitudes toward AI. Students high in neuroticism and agreeableness had more negative attitudes toward AI. Conscientiousness was also linked to negative AI attitudes, while extraversion showed no significant effect.	AI-related content should be integrated into nursing curricula to enhance students’ acceptance. Personality traits should be considered when designing AI training programs. Addressing concerns about AI replacing human roles could improve student perceptions.	Study examined AI attitudes and personality traits; no specific AI tools were used.
Bland (2025)	USA	To enhance medical student engagement and knowledge retention in pharmacology through a multimodal generative AI-based approach using Cinematic Clinical Narratives (CCNs).	40 first-year medical students (18 responded to the Situational Interest Survey)	Mixed-methods study using a multimodal generative AI-driven CCN called ‘Shattered Slippers’, incorporating AI-generated storytelling, images, narration, and theme music.	14/18 students preferred CCN over traditional case-based learning. Students scored an average of 88% on exam questions related to CCN content. Thematic analysis indicated increased engagement, improved recall, and enhanced relatability through pop culture references.	Expand the use of CCNs in medical education. Incorporate AI-driven storytelling to enhance engagement. Develop guidelines for integrating generative AI ethically into medical education.	GPT-4 for storytelling, Leonardo.ai and Stable Diffusion for image generation, Eleven Labs for narration, Suno for theme music composition.
Civaner et al. (2022)	Turkey	To examine the perceptions of medical students on the potential influences of AI on medicine and determine the needs for curriculum restructuring.	3018 medical students from 67 medical faculties	Cross-sectional multi-center study using an online survey distributed nationwide among Turkish-speaking medical students.	85.8% believed AI could facilitate physicians’ access to information, and 76.7% believed it could enhance patient care. 70.5% thought AI could reduce errors in medical practice. 58.6% felt AI might devalue the medical profession, and 45.5% worried about its impact on trust in patient-physician relationships. Only 6.0% felt competent to inform patients about AI applications.	Medical curricula should be updated to include AI-related knowledge and skills. Training on AI applications, reducing medical errors, and addressing ethical concerns should be incorporated into education. A structured AI education framework is needed to prepare future physicians for AI-driven healthcare.	Study examined AI perceptions in medical education; no specific AI tools were implemented.
Ghosh and Bir (2023)	India	To evaluate ChatGPT's ability to solve higher-order competency-based medical education (CBME) questions in medical biochemistry.	200 medical biochemistry reasoning and MCQ-based questions	Cross-sectional study conducted online using ChatGPT (March 14, 2023 version). Questions were selected from an institutional question bank and classified under CBME competency modules.	ChatGPT answered 200 higher-order questions with a median score of 4.0 out of 5. Scores varied across different biochemistry competency modules, with the lowest in oncogenesis and immunity. The inter-rater reliability was outstanding (ICC = 0.926). ChatGPT performed well in traditional concepts but showed limitations in justifying recent advances.	Medical schools should integrate AI-based tools like ChatGPT into education while training students on critical AI evaluation. Continuous updates in AI databases are required to improve responses, particularly in emerging fields like oncology and immunology. Future studies should explore AI's effectiveness across various medical disciplines.	ChatGPT (March 14, 2023 version) for medical biochemistry competency-based question solving.
Lane et al. (2024)	USA	To explore the implementation of generative AI in nursing education and provide a strategic guide for its responsible use.	95 nurse educators from the southeastern United States	Case-based descriptive study using a thematic SWOT (Strengths, Weaknesses, Opportunities, and Threats) analysis of participant responses.	Generative AI was perceived as both a ‘tool and tyrant’. It was identified as a time saver and thief, heightening and dampening innovation, facilitating and impeding critical thinking, and both benefiting and hindering routine tasks.	Establish ethical guidelines for AI use in nursing education. Foster AI literacy among faculty and students. Encourage responsible AI implementation to balance efficiency with ethical concerns.	ChatGPT, Microsoft CoPilot, and other generative AI tools for content creation and educational support.
Rony et al. (2024b)	Bangladesh	To investigate healthcare workers’ knowledge and attitudes regarding artificial intelligence adoption in healthcare.	431 healthcare workers in Dhaka, Bangladesh	Cross-sectional study with stratified random sampling and self-administered surveys. Data analysis included descriptive and inferential statistics, Fisher's exact tests, logistic regression, and Pearson correlation.	Younger age groups (18–25, 26–35) had higher odds of good AI knowledge and positive attitudes. Physicians, full-time employees, and hospital workers exhibited higher knowledge levels. Attending AI conferences and reading research articles were positively associated with AI knowledge. There was a strong positive correlation (r = .89, P < .001) between knowledge and positive attitudes toward AI.	Healthcare training should integrate AI education to bridge knowledge gaps. Targeted AI literacy programs should be developed for different healthcare professional groups. Policies and training must address ethical considerations and job security concerns.	Study focused on AI adoption in healthcare; no specific AI tools were implemented.
Cho and Seo (2024)	South Korea	To examine the dual mediating effects of anxiety and acceptance attitude of AI on the relationship between perception of AI and intention to use AI among nursing students.	180 nursing students from Gyeonggi-do, South Korea	Descriptive cross-sectional study using self-reported questionnaires. Statistical analyses included t-tests, ANOVA, Pearson's correlation, and Hayes's PROCESS macro for mediation analysis.	AI perception positively correlated with acceptance attitude (r = .44, p < .001) and intention to use AI (r = .38, p < .001), but negatively correlated with anxiety (r = −.27, p < .001). Anxiety about AI negatively correlated with both acceptance attitude (r = −.36, p < .001) and intention to use AI (r = −.28, p < .001). Acceptance attitude toward AI had a positive correlation with intention to use AI (r = .43, p < .001). Anxiety and acceptance attitude toward AI had a dual mediating effect on the relationship between AI perception and intention to use AI.	Develop systematic AI-related educational programs to enhance AI perception and competence among nursing students. Address AI-related anxiety through structured training and ethical discussions. Foster a positive AI acceptance attitude to increase its adoption in nursing practice.	Study focused on AI perception and attitudes; no specific AI tools were implemented.
Alruwaili et al. (2024)	Saudi Arabia	To assess nurses’ awareness and attitudes toward AI-integrated tools used in clinical practice.	220 registered nurses from three governmental hospitals in Saudi Arabia	Descriptive cross-sectional study using an online questionnaire administered over four months. Survey included demographic information, AI knowledge assessment, and the General Attitudes toward AI Scale.	70.9% of nurses had basic AI knowledge, while 58.2% were considered ‘aware’ as they had interacted with AI healthcare applications. Nurses expressed moderate openness to AI (Mean = 3.51) but also had concerns regarding AI integration. Significant differences in AI attitudes were observed based on gender, age, and educational background.	Increase AI awareness and training programs for nurses. Address nurses’ concerns about AI's impact on their profession. Develop change management strategies to facilitate AI integration in healthcare.	Study focused on AI awareness and attitudes; no specific AI tools were used.
Wang et al. (2024)	China	To explore the knowledge, attitudes, and concerns of healthcare professionals, AI-related professionals, and others in China toward AI in nursing.	1,243 participants from 25 provinces and municipalities	Cross-sectional online survey conducted between March and April 2024. Questionnaire contained 21 questions across four sections.	57% of participants had little knowledge of AI, and 4.7% had no knowledge at all. 64.7% had little knowledge of AI in nursing, while 13.4% had none. Over 50% of participants had a positive attitude toward AI in nursing. 95.7% believed that strengthening medical ethics toward AI in nursing is necessary.	Enhance AI-related training programs for nursing professionals. Integrate AI education into medical and nursing curricula. Develop ethical frameworks to address concerns about AI in nursing.	Study focused on AI perceptions in nursing; no specific AI tools were implemented.
Moldt et al. (2023)	Germany	To explore medical students’ knowledge and attitudes toward AI and medical chatbots and assess their willingness to integrate these technologies into future medical practice.	12 medical students from the University of Luebeck and the University Hospital of Tuebingen	Mixed-method study using standardized quantitative questionnaires and qualitative group discussions. Data analysis included descriptive statistics and thematic analysis of qualitative responses.	Students generally had a positive attitude toward AI and chatbots but expressed concerns about privacy and data security. 83.3% supported AI for administrative tasks, and 91.7% supported AI in research with health data. 33.3% were concerned about insufficient data protection, and 58.3% feared increased workplace monitoring.	AI and data competencies should be taught in a structured way in medical curricula. Medical schools should incorporate AI ethics discussions to address privacy and surveillance concerns. Future physicians should be trained to critically assess AI applications in clinical practice.	Medical chatbots and AI-based decision support systems.
Labrague et al. (2023)	Philippines, USA	To investigate the attitudes and intentions of student nurses toward Artificial Intelligence (AI) in nursing practice and examine the relationship between their perceptions of AI utilization and their intention to adopt AI technology.	200 student nurses from two government-owned nursing schools	Cross-sectional study using a survey based on the Technology Acceptance Model (TAM). Mediation testing was conducted using Hayes’ PROCESS macro in SPSS (Model 4).	Perceived AI utilization had a significant positive effect on student nurses’ attitudes toward AI (β = 0.450, p < .001) and their intention to adopt AI technology (β = 0.458, p < .001). Attitudes toward AI partially mediated the relationship between perceived AI utilization and the intention to adopt AI (β = 0.255).	Nursing education programs should integrate AI coursework, training, and experiential learning. Healthcare institutions should create supportive environments for nursing students to explore and embrace AI technology.	Study examined AI perceptions in nursing practice; no specific AI tools were implemented.
Allam et al. (2024)	Nine Arab countries (Libya, Egypt, Iraq, Jordan, Syria, Sudan, Algeria, Palestine, and Yemen)	To assess undergraduate medical students’ knowledge, attitudes, and perceptions regarding AI in medicine, with a specific focus on its application in radiology.	4,492 medical students	Multi-national, multi-center cross-sectional study using an online web-based questionnaire. Logistic regression and cluster analysis were conducted to identify predictors and shared response patterns.	92.4% of students had not received formal AI training, and 87.1% exhibited a low level of AI knowledge. 84.9% believed AI would revolutionize medicine and radiology, while 48.9% agreed that it could reduce the need for radiologists. Students with high AI knowledge were more likely to believe AI would replace radiologists and reduce their numbers. Most students agreed AI would aid in automated pathology detection and diagnosis.	Widespread AI education and training should be integrated into medical curricula. Medical schools should focus on addressing AI misconceptions and ethical concerns. Collaboration between AI researchers and educators is needed to optimize AI applications in medicine.	Study focused on AI perceptions in medicine; no specific AI tools were implemented.
Çalışkan et al. (2022)	Turkey	To determine the competencies required for medical graduates to be prepared for AI technologies in medicine and to develop a consensus among experts on these competencies.	94 expert panel members (healthcare professionals, AI researchers, legal/ethics professionals, and medical students)	Three-round e-Delphi study conducted between February 2020 and November 2020. Participants rated AI-related competencies using a 7-point Likert scale.	Strong agreement was reached on 23 AI competencies, including ethical AI use, AI-assisted decision-making, and AI data handling. Four competencies had weaker consensus, such as explaining AI training methods and its strengths/weaknesses.	Medical curricula should integrate AI competency-based training. Future research should explore AI competencies in diverse healthcare settings. AI education should incorporate ethical, technical, and application-based knowledge.	Study focused on AI competencies in medical education; no specific AI tools were implemented.
Karaca et al. (2021)	Turkey	To develop and validate a psychometric scale (MAIRS-MS) for assessing the perceived readiness of medical students regarding artificial intelligence applications in medicine.	568 students (EFA phase), 329 students (CFA phase) from two public universities	Sequential exploratory mixed-method study. Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) were performed to validate the scale.	A four-factor model (cognition, ability, vision, ethics) explaining 50.9% cumulative variance was validated. Cronbach's alpha reliability coefficient was 0.87, confirming strong internal consistency. Medical AI readiness scale (MAIRS-MS) was determined as a valid and reliable tool.	Medical schools should integrate AI literacy and training based on MAIRS-MS findings. The scale can be used for curriculum development and assessing AI readiness among medical students.	Study focused on AI readiness assessment; no specific AI tools were implemented.
Weidener and Fischer (2024)	Germany, Austria, Switzerland	To explore how medical students perceive the use of AI in medicine, as well as the teaching of AI and AI ethics in medical education.	487 medical students from Germany, Austria, and Switzerland	Cross-sectional study using a web-based survey containing 53 items across 6 sections. Statistical analyses included chi-square and Mann-Whitney U tests.	38.8% of students had prior experience with AI-based chat applications (e.g., ChatGPT). 71.7% believed AI would positively impact medicine. 74.9% agreed on the need for AI and AI ethics education, but current offerings were deemed inadequate.	Integrate AI and AI ethics into medical curricula. Enhance AI literacy to prepare students for AI integration in healthcare.	AI-based chat applications such as ChatGPT, Bard, Bing Chat, and Jasper Chat.
Pucchio et al. (2022)	Canada	Examine undergraduate medical students’ perceptions of AI, educational opportunities, and preferred curriculum delivery methods.	486 responses from 17 Canadian medical schools	Cross-sectional mixed-methods study using a survey (32 questions) and interviews with 18 students; Likert-scale responses analyzed quantitatively and qualitatively.	CGAI adoption has grown exponentially since 2019, especially in computer science (32%), healthcare (17%), and business (6%). Top applications include course content generation, research support, and academic writing. Challenges include misinformation, plagiarism risks, privacy concerns, and potential overreliance on AI.	Develop discipline-based AI detection tools to prevent academic abuse. Formulate policies to regulate AI integration in education. Enhance AI literacy among educators and students.	ChatGPT, OpenAI models, VOSviewer, Bibliometrix (R), big data analytics.
Abou Hashish and Alnajjar (2024)	Saudi Arabia	To assess the perceived knowledge, attitudes, and skills of nursing students regarding digital transformation, digital health literacy (DHL), and attitudes toward artificial intelligence (AI).	266 nursing students	Descriptive correlational study using a structured questionnaire covering personal information, knowledge, skills, attitudes toward digital transformation, digital skills, DHL, and attitudes toward AI. Statistical analysis included descriptive statistics and Pearson correlation.	94% agreed AI will be common in medicine; 85% reported no formal AI education; 67% believe AI should be formally taught; 77% think AI optimizes physicians’ work; students lack formal training despite AI's increasing role.	Medical schools should integrate AI into the curriculum; workshops, lectures, and interdisciplinary collaboration preferred; focus on AI's clinical applications, ethical considerations, and literacy.	AI literacy and applications in medicine; image-based diagnostics, workflow optimization, and clinical decision support systems.
Banerjee et al. (2021)	United Kingdom	To evaluate the impact of AI technologies on clinical education from the perspective of postgraduate trainee doctors in London and provide recommendations for trainers.	210 postgraduate trainee doctors in London, UK	Cross-sectional survey conducted between October and December 2020 at NHS postgraduate training centers in London. Survey analyzed with Fisher's exact test, K-modes clustering, and thematic analysis of free-text responses.	58% of trainees perceived an overall positive impact of AI on clinical education. AI was seen as reducing clinical workload (62%) and improving research and audit training (68%). Skepticism was noted in AI's impact on clinical judgement (46% agreement) and practical skills (32%). 92% reported insufficient AI training in their curricula, and 81% supported more formal AI training.	Medical curricula should integrate “Applied AI” topics, including AI ethics, data science, and clinical AI decision-making. AI-assisted education should be leveraged to improve training delivery while ensuring clinical judgement and practical skills are maintained.	AI-based decision support systems, natural language processing for documentation, computer vision for radiology.
Sommer et al. (2024)	Germany	To investigate nurses’ knowledge, perceptions, and experiences with AI in nursing practice in Germany.	114 nurses from various care facilities in Bavaria, Germany	Cross-sectional online survey conducted in June 2023. Convenience sampling method used; data analyzed using descriptive statistics and template analysis for qualitative responses.	Only 25.2% of nurses considered themselves “AI connoisseurs” with substantial AI knowledge. 65.7% viewed AI as an opportunity for nursing, while 13.7% saw it as a threat. Nurses were most familiar with AI applications in patient monitoring (55.7%), route planning (47.7%), and nursing documentation (43.2%). Concerns included job replacement fears, AI's impersonal nature, and high implementation costs.	Increase AI education and awareness in nursing programs. Develop AI policies that ensure ethical use and job security. Encourage user-friendly AI solutions tailored for nursing applications.	Study focused on AI perceptions in nursing; no specific AI tools were implemented.
Creighton et al. (2025)	UK, Australia	To evaluate the effectiveness of a digital serious game in improving nursing students’ self-efficacy regarding academic integrity, academic offenses, professionalism, and artificial intelligence use.	303 first-year nursing students	Quasi-experimental pre-test/post-test design using a bespoke questionnaire. The questionnaire assessed self-efficacy before and after playing the Academic Integrity digital serious game.	Statistically significant improvements were observed across all measured domains, including academic integrity, academic offenses, professional values, feedback utilization, and AI use in academic work. AI use and feedback showed the highest improvements.	Integrate serious games into nursing curricula to enhance ethical learning. Provide structured AI literacy training to nursing students. Encourage interactive digital tools to strengthen academic and professional integrity.	Serious game incorporating AI-assisted scenarios for academic integrity and professional ethics training.

Quality Assessment

In this systematic review, the quality of each included study was meticulously assessed using the CASP checklists to ensure the reliability and validity of the findings (Table 3) (Long et al., 2020). Each author played a significant role in evaluating key aspects of the research methodologies. The process began with an assessment of validity, examining whether the research questions were well-defined and if the study design is appropriately aligned with them. The selection of participants was analyzed for potential biases to maintain the integrity of the outcomes. Reliability was then assessed by evaluating the consistency and reproducibility of data collection methods, ensuring that findings could be replicated in future research.

Table 3.

Quality Assessment of Studies Using CASP Checklists.

Study ID	Research aim Clearly Stated	Appropriate Study Design	Recruitment Methodology Clearly Stated	Ethical Considerations Addressed	Data Collection Methods Clearly Described	Validity of Results	Reproducibility of Study	Applicability to Broader Context	Limitations Discussed
Wood et al. (2021)	Yes	Yes	Yes	Yes	Yes	High	Moderate	High	Yes
Sridharan and Sequeira (2024)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Jackson et al. (2024)	Yes	Yes	Yes	Yes	Yes	High	Moderate	High	Yes
Sami et al. (2025)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Kidwai et al. (2025)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	Moderate	Yes
Rony et al. (2024a)	Yes	Yes	Yes	Yes	Yes	High	High	High	Yes
Issa et al. (2024)	Yes	Yes	Yes	Yes	Yes	High	High	High	Yes
Blease et al. (2022)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	Moderate	Yes
Salem et al. (2024)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Bland (2025)	Yes	Yes	Yes	Yes	Yes	High	High	High	Yes
Civaner et al. (2022)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	Moderate	Yes
Ghosh and Bir (2023)	Yes	Yes	Yes	Yes	Yes	High	Moderate	Moderate	Yes
Lane et al. (2024)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	High	Yes
Rony et al. (2024b)	Yes	Yes	Yes	Yes	Yes	High	High	High	Yes
Cho and Seo (2024)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Alruwaili et al. (2024)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	High	Yes
Wang et al. (2024)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Moldt et al. (2023)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	High	Yes
Labrague et al. (2023)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Allam et al. (2024)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Çalışkan et al. (2022)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	High	Yes
Karaca et al. (2021)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Weidener and Fischer (2024)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	High	Yes
Pucchio et al. (2022)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Abou Hashish and Alnajjar (2024)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Banerjee et al. (2021)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	Moderate	Yes
Sommer et al. (2024)	Yes	Yes	Yes	Yes	Yes	High	High	Moderate	Yes
Creighton et al. (2025)	Yes	Yes	Yes	Yes	Yes	Moderate	Moderate	High	Yes

Additionally, the specificity of the results was scrutinized by analyzing the sensitivity and appropriateness of the measurement tools to confirm that the outcomes directly addressed the research questions. The significance of the findings was also considered, particularly their practical implications and contributions to nursing education. Finally, the applicability of the results was evaluated to determine whether they could be generalized to broader contexts and implemented in real-world educational settings. Overall, most of the included studies were scored as high quality, while others were assessed as moderate quality. This appraisal process enhanced transparency and reproducibility, providing a clear framework for evaluating study rigor. Throughout this process, the collaborative efforts of the authors ensured a systematic and unbiased evaluation, ultimately enhancing the validity and impact of the systematic review's conclusions.

Data Synthesis

The data synthesis approach employed a narrative synthesis, which is well-suited for systematic reviews that include diverse study designs and methodologies. This method enables detailed descriptions and discussions of findings in a narrative format, making it easier to interpret complex and heterogeneous data. Additionally, the synthesis accounted for variations and inconsistencies across studies, ensuring that the conclusions drawn were both robust and representative of the potential state of AI applications in nursing education. To guide the process, this review followed the framework proposed by Popay et al. (2006) for narrative synthesis. Studies were grouped thematically according to their primary focus areas, such as personalized learning, simulation-based training, automated assessment, clinical reasoning, administrative applications, and ethical considerations. Within each theme, findings were compared and contrasted to identify consistencies, divergences, and unique contributions. Meta-analysis was not feasible due to heterogeneity in outcome measures, AI modalities, and research methodologies across the included studies. Therefore, the narrative synthesis framework provided a flexible yet rigorous approach to organize and interpret results.

To ensure robustness, the researchers performed sensitivity analyses by considering study design, methodological quality (based on CASP assessments), and risk of bias. Findings from studies assessed as low quality were interpreted cautiously and compared against higher-quality evidence to evaluate stability of conclusions. Although quantitative meta-analysis was not possible, potential publication bias was considered qualitatively. This involved assessing the distribution of included studies across countries, publication outlets, and study types, and reflecting on whether overrepresentation of positive findings may have influenced conclusions.

Results

Overview of Study Selection

The study selection process for this systematic review followed a structured approach to ensure the inclusion of relevant and high-quality research. Initially, a total of 582 records were identified through comprehensive searches across multiple databases, including 200 from PubMed, 120 from CINAHL, 150 from IEEE Xplore, and 112 from Scopus. After removing 82 duplicate records, the remaining 500 studies were subjected to a title and abstract screening process. Based on predefined inclusion and exclusion criteria, 320 articles were excluded at this stage for reasons such as lack of relevance in 140 studies, poor study design in 110 studies, or lack of outcome measures in 70 studies. Following the title and abstract screening, 180 full-text articles were sought for retrieval. However, 15 studies could not be retrieved, leaving 165 studies for full-text assessment of eligibility. Studies were excluded from being non-peer-reviewed in 22 cases, having poor methodology in 28 cases, small sample sizes in 18 cases, irrelevance in 30 cases, data bias in 22 cases, or lack of clear outcome measures in 17 cases. After the full-text review, a total of 28 studies met the inclusion criteria and were included in the qualitative synthesis. The included studies comprised different research designs, with cross-sectional studies being the most common (18/28; 64.3%). Qualitative studies accounted for 4/28 (14.3%), mixed-methods studies also represented 4/28 (14.3%), while Delphi and quasi-experimental studies each contributed 1/28 (3.6%).

Characteristics of Included Studies

The included studies, conducted in the USA, India, Pakistan, the Netherlands, and Bahrain, explored the role of AI in nursing education using diverse methodologies such as cross-sectional surveys, semi-structured interviews, online questionnaires, proof-of-concept studies, and mixed-method approaches. Sample sizes varied, with some studies involving large cohorts of 702 and 325 students, while others focused on smaller groups, such as 15 students in qualitative interviews. A range of AI technologies was examined, including general-purpose large language models (LLMs) such as those developed by OpenAI (e.g., ChatGPT), Anthropic (Claude), Google (Gemini), and Meta, as well as machine learning (ML) applications. In cases where studies did not specify exact versions or platforms, these tools were treated collectively as general-purpose LLMs/ML classifiers. Some studies assessed AI literacy and perceptions without specifying particular AI tools, whereas others explored AI applications in structured learning objective generation, AI-assisted curriculum development, and automated medical training support.

Findings suggest that while AI holds significant promise in nursing education, its awareness and adoption remain inconsistent among students and faculty. While some research highlights high acceptance rates and AI's potential to enhance learning efficiency and decision-making, others report limited AI proficiency among educators and healthcare professionals. To bridge this gap, recommendations emphasize integrating AI into nursing curricula, developing AI-specific educational tools, incorporating ML training, and optimizing AI-driven learning approaches to improve education quality, clinical decision-making, and adaptive learning experiences. However, most included studies were concentrated in middle- and high-income countries, which may limit the generalizability of findings to resource-limited settings. Equity considerations are important, as technological infrastructure, faculty preparedness, and access to AI-driven tools vary widely across contexts, suggesting that benefits observed in well-resourced environments may not be directly transferable to low-resource educational systems.

Results: Integration of AI in Nursing Education

High-Acuity Settings: Simulation-Based Learning and Decision Support

AI-driven simulations in high-acuity environments replicate emergency conditions such as trauma resuscitation, cardiac arrests, or respiratory failure using virtual and augmented reality platforms (Issa et al., 2024; Salem et al., 2024). These immersive tools allow learners to engage in complex clinical situations with instant, AI-generated feedback tailored to their performance level (Bland, 2025). The real-time adaptation of difficulty ensures that learners remain engaged and challenged, while instructors use performance dashboards to assess competency and conduct targeted debriefings (Civaner et al., 2022). Blease et al. (2022) found that such platforms improve clinical judgment and emotional readiness. In one pre/post-test study, students exposed to AI simulations showed a statistically significant improvement in clinical confidence with a mean gain of 1.06 on a 5-point scale (Creighton et al., 2025). AI simulations in high-acuity training enhance emergency preparedness by providing risk-free, personalized, real-time learning experiences.

Ambulatory Settings: AI-Supported Clinical Reasoning

AI tools in ambulatory settings focus on enhancing diagnostic reasoning through longitudinal, case-based simulations. These platforms present evolving patient scenarios enriched by AI-generated diagnostic suggestions, which help students interpret lab results, assess symptoms, and compare intervention outcomes (Labrague et al., 2023; Çalışkan et al., 2022). Machine learning algorithms predict patient trajectories and simulate outpatient decision-making workflows, often embedded in electronic health records (Moldt et al., 2023). These tools promote anticipatory thinking by presenting learners with evolving clinical cases. Sridharan and Sequeira (2024) highlight that AI modules help bridge the gap in clinical supervision for students during ambulatory care rotations. AI-driven case simulations in outpatient contexts sharpen diagnostic skills and encourage predictive, patient-centered reasoning.

Academic Settings: Personalized Learning Platforms

In academic environments, AI is primarily used to customize learning paths. These adaptive systems track individual progress, analyze performance data, and deliver tailored content based on knowledge gaps (Wood et al., 2021; Sami et al., 2025). Features include dynamic quiz generation, personalized study recommendations, and just-in-time remediation (Ghosh & Bir, 2023). AI tutors replicate conversational engagement and simulate diagnostic pathways using virtual patient interactions (Kidwai et al., 2025; Rony et al., 2024a). Gamified elements such as achievement tracking and competitive leaderboards further enhance engagement (Jackson et al., 2024). These platforms also provide predictive analytics to identify students at risk and recommend support interventions (Allam et al., 2023; Sridharan & Sequeira, 2024). AI-powered personalization supports learner autonomy and improves outcomes through targeted, data-informed instructional delivery.

Assessment and Feedback: Automated Evaluation Systems

Automated assessment systems use AI to evaluate clinical reasoning, communication, and procedural knowledge. Tools employing natural language processing analyze student responses in reflections, quizzes, and simulated conversations (Cho & Seo, 2024; Lane et al., 2024). These systems reduce grading time and eliminate subjective bias, improving the consistency of evaluation across cohorts (Civaner et al., 2022). Feedback loops are central to these platforms. AI systems provide real-time performance reports during simulations, highlighting strengths and pinpointing reasoning errors (Rony et al., 2024b; Wang et al., 2024). These immediate responses reinforce learning and reduce the likelihood of repeating mistakes. Longitudinal tracking also allows educators to monitor learner growth and address knowledge gaps proactively (Alruwaili et al., 2024). Karaca et al. (2021) underscore the effectiveness of these systems in fostering reflective learning and academic self-regulation. AI-enabled assessments deliver rapid, consistent feedback that accelerates clinical reasoning and performance improvement.

Institutional Applications: Administrative and Curriculum Management

At the institutional level, AI systems streamline academic administration and curriculum design. Learning management systems enhanced with AI can track attendance, assignment submissions, and engagement metrics to identify students at risk (Pucchio et al., 2022; Sommer et al., 2024). Predictive models assist administrators in making data-driven decisions about scheduling, resource allocation, and curriculum development (Wang et al., 2024). AI-based chatbots handle common administrative inquiries, improving responsiveness and reducing staff workload (Abou Hashish & Alnajjar, 2024; Banerjee et al., 2021). These systems also enable competency-based education by allowing students to progress according to skill mastery rather than fixed timelines (Creighton et al., 2025; Rony et al., 2024b; see Table 4 for comparative details). Despite the promise, barriers remain. Implementation requires significant infrastructure investment and faculty training (Cho & Seo, 2024; Kidwai et al., 2025). Faculty engagement is crucial to ensure alignment between AI-generated insights and pedagogical goals (Jackson et al., 2024; Lane et al., 2024). AI integration into institutional systems supports efficient academic planning, student monitoring, and personalized support at scale.

Table 4.

Comparative Summary of AI Applications in Nursing Education.

Study/Author(s)	Modality	Comparator	Primary Outcome	Effect Direction/Magnitude	Implementation Notes
Creighton et al. (2025)	AI Simulations (VR)	Traditional training	Clinical confidence (↑1.06 scale)	Positive; statistically significant	Requires VR setup; faculty training needed
Wood et al. (2021)	Adaptive Learning Platform	Standard e-learning	Student performance improvement	Moderate positive	Integrated easily into LMS
Jackson et al. (2024)	Gamified AI tutor	Static module	Engagement and knowledge retention	High engagement boost	Some faculty tech resistance
Civaner et al. (2022)	AI Assessment Feedback	Manual grading	Evaluation speed and consistency	Strong improvement in grading time	Privacy policies needed
Pucchio et al. (2022)	AI Curriculum Dashboard	Manual monitoring	Academic risk detection	Predictive accuracy increased	Training for admin staff essential
Alruwaili et al. (2024)	NLP Feedback System	Instructor-only feedback	Error correction and learner reflection	Effective in simulation debriefs	Requires broadband access and NLP language models
Rony et al. (2024b)	Virtual AI Tutor	Video-based learning	Personalized learning	High adaptability, student satisfaction	Ongoing content updates required
Moldt et al. (2023)	ML Diagnostic Support	Human-only reasoning	Diagnostic accuracy	Moderate increase in complex case accuracy	Relies on structured case database
Banerjee et al. (2021)	Chatbot for Admin Support	Email-based support	Response time and student satisfaction	Reduced burden on staff	Language customization important
Bland (2025)	VR Clinical Simulation	Static scenario training	Decision-making confidence	Improved realism and stress response	Limited access in low-resource settings

Cross-Setting Considerations: Ethics and Implementation Challenges

While the integration of AI has advanced educational outcomes, ethical concerns remain. Data privacy, informed consent, and algorithmic transparency are major issues as AI systems handle sensitive student information (Civaner et al., 2022; Karaca et al., 2021). Jackson et al. (2024) and Blease et al. (2022) caution against over-reliance on AI, which may diminish the role of human mentorship and affect the development of empathy and interpersonal skills. Educators also expressed concern over feedback generated by AI systems that sometimes contradict human instructors or appear impersonal (Alruwaili et al., 2024; Weidener & Fischer, 2024). Implementation inequities are evident in under-resourced settings, where infrastructure gaps and poor connectivity restrict AI access (Abou Hashish & Alnajjar, 2024; Sommer et al., 2024). Faculty readiness remains a critical bottleneck. Many instructors lack the technical training needed to implement AI tools effectively, limiting their pedagogical potential (Bland, 2025; Moldt et al., 2023). Ethical AI adoption in nursing education requires faculty training, transparency, and equitable access to technological infrastructure.

Discussion

This systematic review demonstrates how AI applications can either augment or complicate nursing education depending on their alignment with educational contexts, workflows, and stakeholder readiness. In environments where AI systems were purposefully integrated to support existing pedagogical strategies (such as adaptive learning tools mirroring individualized instruction or simulation platforms reflecting real clinical encounters), the benefits were more pronounced (Marr & Tsang, 2025). These outcomes often stemmed from congruence between AI capabilities and instructional needs, fostering greater engagement, efficiency, and diagnostic acumen (Akca Sumengen et al., 2025; Huang et al., 2025). In contrast, implementation in contexts lacking institutional preparedness, faculty buy-in, or technological infrastructure tended to yield modest or uneven results (Alwadani et al., 2024; Habib et al., 2024; Simms, 2025).

One key driver of successful implementation was the integration of AI into student-centered workflows that promoted autonomy and personalization. When students could control the pace and depth of their learning through real-time feedback and interactive modules, AI became a scaffold for critical thinking rather than a prescriptive tool (Sridharan & Sequeira, 2024; Sami et al., 2025). Conversely, resistance emerged when AI tools were perceived as undermining human mentorship or oversimplifying complex clinical reasoning tasks. This highlighted a trade-off between automation and the cultivation of nuanced judgment (Bodur et al., 2024; Fazlollahi et al., 2022). Educators also expressed concerns about the erosion of interpersonal skills and ethical reasoning, which remain core competencies in nursing practice and are difficult to capture through algorithmic assessments (Harishbhai Tilala et al., 2024; Russell et al., 2023).

Trust and accountability emerged as central boundary conditions in the integration of AI. Faculty skepticism often stemmed from limited AI literacy and concerns about data privacy, fairness, and opaque algorithmic processes (Alghamdi & Alashban, 2024; Alwadani et al., 2024). These concerns were exacerbated in systems where AI decisions were not easily auditable or lacked transparent justifications (Alier et al., 2021; De Gagne et al., 2024). Similarly, students demonstrated reluctance when feedback from AI lacked contextual nuance or appeared inconsistent with human evaluators. To mitigate these tensions, studies emphasized the need for a “human-in-the-loop” model. This approach embeds human oversight into automated systems to balance efficiency with ethical safeguards (Jonathan, 2025; Lane et al., 2024; Pucchio et al., 2022). It also fosters greater trust in AI-enhanced learning environments, particularly when systems are designed to augment (rather than supplant) the educator's role.

Translating these insights into actionable principles, the review suggests several implementation strategies. Aligning AI adoption with pedagogical goals is essential, ensuring that tools complement rather than disrupt faculty workflows (Grunhut et al., 2022; Jha et al., 2022). Faculty training should extend beyond technical know-how to include ethical reasoning, data governance, and scenario-based decision-making to maintain accountability (Marr & Tsang, 2025; Mehta et al., 2021). Tool selection should also be guided by equity considerations, making sure that AI does not widen digital divides but instead enhances inclusivity through accessible formats and mobile-friendly designs (Kansal et al., 2022; Lee et al., 2021; Majumder & Haque, 2025). In addition, iterative evaluation should be embedded in implementation plans, with metrics tied to learning outcomes, user satisfaction, and system transparency (Harrison, 2024).

In resource-constrained settings, where infrastructure limitations and workforce shortages pose formidable barriers, the findings offer critical guidance. Phased deployment of AI tools (starting with low-cost, open-source platforms or AI-integrated mobile learning apps) can enable gradual capacity-building without overburdening institutions (Mousavi Baigi et al., 2023). Faculty development initiatives, including regional train-the-trainer models, can multiply impact while minimizing cost (Castonguay et al., 2023). Cloud-based platforms and regional resource-sharing networks could facilitate collaborative curriculum design and AI deployment, especially for simulation and assessment tools (Choi et al., 2025; Farghaly Abdelaliem et al., 2022). These strategies, if guided by community needs and aligned with curricular goals, can help bridge the digital divide and ensure that AI contributes meaningfully to equitable nursing education across diverse global contexts.

Implications for Education

To support structured AI integration in nursing curricula, this study proposes a four-tier competency ladder: Awareness, User, Supervisor, and Designer. At the Awareness level, learners should recognize AI's role in healthcare and education, with outcomes such as identifying AI tools and articulating ethical concerns. Assessment may include a multiple-choice quiz or reflective paragraph. Suggested activity: completing a short module on AI in nursing. The User level focuses on operational use of AI tools in simulations or personalized learning platforms. Outcomes include demonstrating AI-assisted decision-making. Assessment methods include Objective structured clinical examination (OSCE) stations or skill-based rubrics. A suggested activity is engaging in a chatbot-based clinical case. At the Supervisor level, learners critically appraise AI-generated outputs and guide ethical applications. Outcomes include evaluating data bias and explaining human-AI collaboration. Assessment may involve annotated case reviews or portfolio artifacts. A suitable activity is mentoring junior peers in AI-supported labs. At the Designer level, learners co-develop AI-enhanced teaching tools or simulations. Outcomes include creating a use-case or prototype for AI integration. Capstone projects or pilot-tested instructional materials serve as assessments.

To scale this model, institutions should implement a faculty development program of 12–15 h over 4 weeks, covering AI fundamentals, hands-on practice, ethics, and curriculum mapping. Resources needed include interactive demos, sample rubrics, and AI toolkits. Uptake can be evaluated via pre/post-tests, peer-reviewed teaching plans, and self-reported implementation rates. This framework enables rapid yet sustainable AI integration by aligning competencies with teaching strategies, assessments, and institutional support systems.

Implications for Practice

Practice Implementation and Staged Rollout

A staged rollout model is recommended for the safe and effective integration of AI tools into clinical education. This includes three phases: (1) Discovery: identify specific clinical use-cases; responsibility lies with unit heads or nurse educators; success is measured by relevance to workflow and stakeholder readiness. (2) Pilot: implement small-scale trials with run-chart metrics such as error detection rates, time-to-completion of tasks, and reliability of escalation; faculty leads and clinical supervisors oversee this phase. (3) Scale: institutional governance ensures alignment with accreditation standards, while audit trails track AI tool performance, outcomes, and human oversight. A pre-implementation checklist should include workflow mapping, identification of high-risk failure points, ethical risk scenarios, and a defined fallback plan if the AI tool underperforms or fails. Documenting oversight mechanisms is critical to maintain safety and accountability in AI-supported clinical decisions. This model allows nursing institutions to evaluate efficacy, minimize risk, and foster trust while integrating AI into real-world practice.

Ethics and Governance

To support ethical and accountable AI integration in nursing education, this review proposes a practical governance bundle organized around four key principles: transparency, fairness, safety, and responsibility. Transparency involves documenting the model type, version, and rationale for use to promote clarity in AI selection. Fairness includes conducting pre-defined checks that reflect the needs of the local learner population, such as evaluating bias or performance across demographic groups. Safety requires identifying specific trigger conditions for AI activation and establishing escalation pathways for when AI recommendations are uncertain or potentially harmful. Responsibility entails assigning named individuals or teams to oversee monitoring, system updates, and ethical compliance.

For institutional auditability and continuous improvement, this study proposes recommends storing AI prompts, configuration settings, version history, and evaluation rubrics in a shared appendix or repository accessible to faculty and governance committees. This centralized documentation will support periodic review, ensure transparency, and strengthen trust in AI tools used for education. Implementing this governance approach will enable nursing institutions to integrate AI responsibly while safeguarding student learning and ethical standards.

Practice-Oriented Recommendations and Future Directions

Based on the review findings, the researchers propose three testable recommendations to guide future research and practice. First, evaluate the impact of AI-based simulation tools on clinical reasoning using a randomized design, with competency scores as the primary endpoint assessed over a 6-month follow-up. Second, assess AI-driven personalized learning platforms via a quasi-experimental design, focusing on knowledge retention and skill application at 3 months. Third, examine ethical AI integration by implementing a faculty development intervention, measuring changes in AI literacy and ethical confidence over a 12-week period. To enhance cross-study comparability, this review advocates for a minimal common outcome set including measures such as learner engagement, diagnostic accuracy, faculty workload, and patient safety proxies. Implementation reporting should include training time, required staffing, and downtime protocols. Finally, this study urges replication of interventions across high- and low-resource institutions to assess scalability, feasibility, and equity of AI adoption in nursing education (Table 5).

Table 5.

Prioritized Recommendations for Future Practice and Research.

Priority Area	Specific Findings From This Review	Recommendation	Measurable Outcome	Suggested Study Design
AI literacy	AI literacy was inconsistent across institutions; students and educators showed limited proficiency despite high perceived usefulness. Targeted frameworks are needed to standardize baseline competencies	Implement an AI literacy framework across prelicensure curricula	% of students achieving competency benchmarks on AI knowledge tests	Multicenter cross-sectional or quasi-experimental study
Simulation efficacy	AI-driven VR simulations improved clinical confidence (mean +1.06 on 5-point scale, p < .05), engagement, and realism in emergency scenarios; effectiveness varied by infrastructure availability	Compare AI-enhanced VR simulations vs. standard simulation training	Objective Structured Clinical Examination (OSCE) performance at 3 months	Multicenter randomized controlled trial (RCT)
Assessment integration	AI-powered assessment tools delivered real-time, unbiased feedback and enhanced grading efficiency, though subjective competencies like empathy required human validation	Evaluate AI-assisted grading tools in nursing education	Reduction in grading turnaround time; inter-rater reliability vs. faculty grading	Cluster RCT or quasi-experimental study
Learning analytics	AI-supported analytics flagged struggling students early and informed adaptive teaching, though challenges remained around ethical data usage and standardization across systems	Define and test a minimum dataset for student learning analytics	Predictive accuracy for at-risk student identification (AUC/ROC)	Prospective cohort study
Faculty training	Faculty lacked readiness to use AI tools confidently; structured training programs improved attitudes and integration skills when aligned with clinical teaching goals	Assess structured AI training programs for nursing faculty	Hours of training completed; improvements in AI teaching self-efficacy scores	Pre-post intervention study
Ethical/governance frameworks	Human-in-the-loop models ensured oversight of AI-generated feedback, preserving empathy and ethical reasoning while reducing faculty workload	Test human-in-the-loop assessment models	Concordance rates between AI feedback and faculty judgment	Mixed-method evaluation
Cost-effectiveness	TCO varied by program needs; simulation labs required greater investment. Resource-limited settings benefited from phased implementation and open-source AI to improve equity	Analyze the economic impact of AI tool adoption	Total cost of ownership (TCO) vs. improvement in student outcomes	Cost-effectiveness analysis

Strengths and Limitations

This review has several limitations. Despite these limitations, a key strength of this review lies in its comprehensive narrative synthesis across diverse educational settings and AI modalities, supported by a rigorous quality appraisal framework. First, there was notable heterogeneity across educational settings, AI modalities, and outcome measures, which limited comparability and prevented meta-analysis. Some outcome defects such as equity, long-term skill retention, and patient-level effects, were insufficiently reported. The review focused exclusively on English-language, peer-reviewed publications, introducing possible selection and publication bias. Additionally, most studies were from high-income countries, limiting generalizability to low-resource settings. Rapid developments in AI models, especially LLMs, may also affect the future applicability of current findings. Although a rigorous quality appraisal was conducted, sparse data in some areas reduced the ability to assess effect durability. To address these concerns, sensitivity checks were applied based on study quality and design, and this study recommends future reviews include grey literature and non-English studies to broaden perspectives. These limitations may tilt interpretation toward optimistic findings and underscore the need for more robust, longitudinal, and inclusive research.

Conclusion

AI is currently most actionable in nursing education through adaptive learning systems, simulation-based training, and automated assessments that support personalized instruction and diagnostic reasoning. For safe and effective integration, governance mechanisms must ensure data privacy, transparency, role-based accountability, and ethical oversight. The first measurable improvement should focus on enhancing diagnostic accuracy among students, as it directly impacts patient care and clinical readiness. A critical next step is to conduct a multi-institutional, longitudinal study that evaluates the effectiveness of AI-enhanced clinical simulations compared to traditional methods. This evaluation should assess performance outcomes, user trust, and implementation feasibility across varied resource settings. Such evidence could drive meaningful policy changes, inform curriculum design, and establish standards for scalable AI adoption in nursing education. By aligning innovation with pedagogical and ethical priorities, stakeholders can responsibly harness AI to improve learning outcomes while preserving the human values at the heart of nursing practice.