https://orcid.org/0009-0001-4610-2075
Figures
Abstract
With the rapid development of artificial intelligence (AI), large language models (LLMs), such as ChatGPT have shown potential in medical education, offering personalized learning experiences. However, this integration raises ethical concerns, including privacy, autonomy, and transparency. This study employed a scoping review methodology, systematically searching relevant literature published between January 2010 and August 31, 2024, across three major databases: PubMed, Embase, and Web of Science. Through rigorous screening, 50 articles which met inclusion criteria were ultimately selected from an initial pool of 1,192 records. During data processing, the Kimi AI tool was utilized to facilitate preliminary literature screening, extraction of key information, and construction of content frameworks. Data reliability was ensured through a stringent cross-verification process whereby two independent researchers validated all AI-generated content against original source materials. The study delineates ethical challenges and opportunities arising from the integration of AI and LLMs into medical education, identifying seven core ethical dimensions: privacy and data security, algorithmic bias, accountability attribution, fairness assurance, technological reliability, application dependency, and patient autonomy. Corresponding mitigation strategies were formulated for each challenge. Future research should prioritize establishing dedicated ethical frameworks and application guidelines for AI in medical education while maintaining sustained attention to the long-term ethical implications of these technologies in healthcare domains.
Citation: Li X, Yan X, Lai H (2025) The ethical challenges in the integration of artificial intelligence and large language models in medical education: A scoping review. PLoS One 20(10): e0333411. https://doi.org/10.1371/journal.pone.0333411
Editor: Ahmed Hosny Saleh Metwally, Helwan University Faculty of Education, EGYPT
Received: March 31, 2025; Accepted: September 12, 2025; Published: October 22, 2025
Copyright: © 2025 Li et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting information files.
Funding: This study was supported by the 2022 National Innovation & Entrepreneurship Training Program for College Students (Grant No. 202210633008X), the Research Program of Sichuan TCM Culture Coordinated Development Research Center (Grant No. 2021XT31), and the 2023 National Traditional Chinese Medicine Examination Research Project (Grant No. TC2023019). In addition to the aforementioned sources, there was no additional external funding received for this study.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Artificial intelligence (AI) originally referred to as the simulation of human intelligence processes through algorithms and data analysis originated in the 1950s [1]. The emergence of ChatGPT in 2022 catalyzed the rapid development of large AI-based language models (LLMs), which shows AI’s rapid growth and progress [2]. AI and LLMs show promising applications in various fields, such as scientific research, statistical analysis, and machine translation, especially in health care and medical education [3–5].
With their powerful data processing and natural language understanding capabilities, AI and LLMs are changing how medical knowledge is acquired, taught, and applied [6].This transformation aligns closely with the core objectives of medical education, which aims not only to impart knowledge but also to cultivate clinical skills, professional attitudes, and ethical decision-making abilities—reflecting its unique focus on shaping medical professional values [7]. A complex and extensive body of knowledge is required for medical students and staff, and AI and LLMs can achieve this goal by integrating and analyzing vast datasets [8]. AI leverages machine learning and computer vision to focus on medical operational skill training in medical education, such as virtual surgical simulation, whereas LLMs utilize their language comprehension and generation capabilities to emphasize knowledge integration and interactive reasoning, primarily driving clinical knowledge synthesis and decision-making exercises, such as the simulation of complex physician-patient communication scenarios through natural language interactions. Studies have shown that AI and LLMs have the potential to enhance the learning experience and quality of education for medical students [9,10].
However, integrating AI and LLMs into medical education also faces ethical challenges [11], which should be contextualized within the established framework of medical ethics principles—namely respect for autonomy, beneficence, non-maleficence, and justice [12]. Patients’ right to know and privacy protection, the accessibility and affordability of technology, and the definition of responsibility in the context of AI-assisted medical education are critical ethical issues that need to be addressed urgently [13–15]. As we stand on the cusp of a new era of medical education, we must address these challenges with foresight and prudence.
Objectives
In this study, we conducted a scoping review to explore the ethical challenges faced in integrating AI and LLMs in medical education. We set two objectives: first, to synthesize existing literature elucidating critical ethical challenges arising from AI and LLM applications in medical education; second, to explore solutions, outline actionable implementation pathways, and establish a foundational discourse for responsibly leveraging these technologies to support evidence-based decision-making and educational practices.By achieving these objectives, we aimed to advance the responsible and effective integration of AI and LLMs into medical education..
Methods
Study design and registration
This scoping review was conducted according to the framework established by Arksey and O’Malley (Arksey et al., 2005) as well as the PRISMA Extension for Scoping Reviews (PRISMA-ScR) [16]. The PRISMA-ScR checklist is shown in S1 File. The study has been pre-registered on the OSF platform (https://doi.org/10.17605/OSF.IO/65TYD).
Search strategy
The researchers systematically searched relevant articles published from 2010 to August 31, 2024 in the PubMed, Embase, and Web of Science databases, with the year 2010 chosen as the starting point to capture breakthrough advancements in deep learning, the rise of structured electronic health data, and the transition of AI applications in healthcare from theoretical frameworks to clinical validation. We used a combination of subject terms and free text terms, and adjusted the search according to the characteristics of each database. The search terms used are listed in Table 1. The complete search strategy can be found in S2 File.
Inclusion and exclusion criteria
Inclusion criteria.
(1) The article discusses the application of the integration of AI and LLMs in medical education, and explores ethical issues in the main text. Studies on non-education-related AI technologies (e.g., genomic bioinformatics analysis) are excluded. (2) The article explicitly mention or implicitly address ethical challenges related to AI implementation (e.g., data privacy, algorithmic bias, accountability disputes), with substantial discussion beyond superficial keyword references. (3) The article was published from January 1, 2010 to August 31, 2024.
Exclusion criteria.
(1) To prioritize the unique demands of physician education,articles that solely focus on non-clinical medical education (e.g., nursing, pharmacy, dentistry) without addressing clinical medical education are excluded;(2) Articles that mention medical education only as a broad contextual element without in-depth analysis of AI implementation processes,or include medical education as a subject matter but do not specifically concentrate on it are excluded.
Study selection and data abstraction
After removing the duplicate studies, two reviewers (Li and Yan) independently screened titles/abstracts and assessed full texts using predefined criteria. A pilot calibration (50 random articles, κ = 0.85) ensured initial consistency. All decisions were validated in duplicate, with disagreements resolved through team discussions and adjudication by a senior researcher (Lai), following PRISMA guidelines.
Collating and summarizing data
Data collection and organization are achieved through a pre-designed table, including the first author, publication date, country, language, journal/periodical, title, document type, research topic, current ethical challenges of integration of AI and large language models in medical education, and suggestions to address these ethical challenges etc. Data collection is completed by two independent researchers, assisted by KIMI AI, who screen and cross-check the data. Disputed content was resolved through group discussion.
Use of AI tools
(1) Names of AI tools used: We utilized Kimi AI to assist in various aspects of our research. (2) Description of how the tools were used: Kimi AI was employed to facilitate the initial screening of literature and extraction of key information, such as authors, publication dates, research topics, and main findings. It also assisted in generating preliminary outlines for sections of the manuscript and suggesting relevant research questions. (3) Evaluation of the validity of the tool’s outputs: To ensure the accuracy and reliability of the information generated by Kimi AI, we implemented a rigorous review process. Two independent researchers cross-checked the AI-generated content against the original sources. Any discrepancies or uncertainties were resolved through group discussions, ensuring that all data and information included in the study were thoroughly validated [17,18]. The complete record file of the use of Kimi AI can be found in S3 File.
Critical appraisal within sources of evidence
The main objective of this review was to offer a comprehensive summary of the current literature and to integrate information for a broader understanding of the subject rather than performing an in-depth critical evaluation of individual studies. As such, we did not carry out a critical assessment of the evidence sources. The decision not to critically appraise the selected articles stemmed from it not being a goal of this review and would not provide any further significant insights [19].
Results
Study selection
We identified 1192 relevant literature records from the database, excluded 426 duplicate records, and excluded 685 literature records by reading the titles and abstracts. Then, we excluded another 30 literature records by reading the full texts, and finally, 50 literature records were included in the study [6,13, 20–66]. The PRISMA flow diagram for this study is shown in Fig 1. All the data are presented in S1Table.
PRISMA flow diagram summarizing the study’s article selection process. A total of 1192 articles were identified. Finally, 50 studies were included.
Characteristics of the sources of evidence
The summary of the included articles in our scoping review highlights the distribution and characteristics of the evidence base. Regarding geographical distribution, most papers were published in China and the United States, with 10 (20.00%) and 7 (14.00%) articles respectively. Other countries making notable contributions include Iran with 5 articles (10.00%), and Austria, Turkey, Australia, Germany, India, and Oman, each contributing 3 articles (6.00%). The distribution of publications across the years in our scoping review reveals a marked increase in scholarly output. There was a mere 1 publication (2.00%) in 2021, which increased to 4 publications (8.00%) in 2022, and significantly peaked at 24 publications (48.00%) in 2023, followed by 21 publications (42.00%) in 2024, indicating a substantial and recent focus on the integration of AI and LLM in medical education. A minority of the articles originated from a diverse range of other countries, each with a single publication, which is not detailed due to their low frequency. English was the predominant language of publication, with 48 articles (96.00%) written in English, followed by Korean with 2 articles (4.00%). This emphasizes the international nature of the discourse on AI and LLM in medical education. When examining the types of articles, a near-equal division was observed between reviews (15, 30.00%) and articles (16, 32.00%). Viewpoints accounted for 7 articles (14.00%), while original papers, commentaries, editorial materials, and preprints contributed a smaller proportion. The remaining categories, including letters, opinions, and perspectives, each accounted for a single article (1.96%). A total of 50 articles were considered, with a significant majority focusing on the application of AI and LLM in medical education, specifically highlighting the potential applications or benefits in 37 articles (74%) and the ethical issues in 13 articles (26.00%). This overview of the characteristics of the evidence sources provides a structured summary of the literature included in our scoping review, offering insights into the current scope and focus of research on AI and LLMs in medical education. The study characteristics are shown in Table 2 and Fig 2. S2 Table is the standardized study characteristics table.
Overview of AI and LLM applications in medical education
In a review of 50 selected articles, AI and LLMs’ most frequently discussed applications in medical education include their ability to provide personalized learning plans and feedback on student’s strengths and weaknesses, thereby enhancing learning efficiency and quality (84%). AI technology also plays a crucial role in simulating patient interactions and clinical scenarios, which supports the improvement of students’ diagnostic and interpersonal skills (75%). Furthermore, AI and LLMs foster interactive learning by providing dynamic, immersive educational experiences that replicate clinical environments (63%). Additionally, AI demonstrates significant value in facilitating simulated dialogues, intelligent tutoring, and automated Q&A functions, all of which aid in problem-based learning (PBL), team-based learning (TBL), case-based learning (CBL), and precision medical education. In research support, AI and LLMs have proven instrumental by automating literature reviews, offering research design recommendations, and conducting statistical analyses, thus contributing substantially to medical research [63–65].
Ethical challenges and recommendations
Our review of the same corpus of literature also revealed a complex web of ethical considerations arising from the integration of AI and LLMs in medical education, as summarized in Fig 3. The following sections provide a detailed qualitative analysis of each major ethical theme.
The bar chart differentiates between two categories: current ethical challenges (represented by pink bars) and recommendations to address these challenges (represented by blue bars).The length of each bar corresponds to the number of studies discussing each item, with the exact count explicitly labeled within each bar for clarity. Current ethical challenges.
Privacy and data security.
Privacy and data security are the most prominent ethical challenges when AI and LLMs are applied in medical education. Given the extensive utilization of these technologies, safeguarding the sensitive health information of both students and patients is of paramount significance. Ensuring data confidentiality and preventing unauthorized access and misuse is a crucial concern [21,22,33,34,41,42,44,46,48,50,54]. The personalization of services and algorithm enhancements by AI and LLMs necessitate vast quantities of personal data, giving rise to issues related to privacy breaches and data security [59]. For instance, Rodriguez Peñaranda N’s study [67] emphasize that AI applications in surgical training, particularly for kidney cancer surgeries, entail extensive data collection, encompassing patient and procedural data. Additionally, the storage and transmission of patient examination data, such as ultrasound images, on AI platforms might expose vulnerabilities, as noted by Koçer Tulgar Y’s study [37]. These challenges are exacerbated by risks such as hacking, system breaches, or insufficient encryption measures, resulting in unauthorized access or data interception during transmission [60]. Data collection, frequently carried out without users’ awareness or explicit consent, presents specific risks when dealing with patient data or student information in educational settings [45,60]. The limited accessibility of regulated datasets and the absence of mechanisms to verify data authenticity further blur the distinctions between legitimate and unethical data usage. As pointed out by Gurnani B [32], AI may not fully comprehend the context of a problem and the subtle distinctions in ophthalmology cases, which could lead to privacy and data security issues.
Transparency.
The lack of transparency in AI decision-making processes, commonly referred to as “black box algorithms,” poses a significant challenge to the application of artificial intelligence and large language models in medical education [22,28]. This issue has been extensively discussed in the included literature, with various studies highlighting the problem of opacity in AI systems from multiple dimensions [22,28,35,38–40,42,45,48,55,57–59,62,66,68]. The core problem lies in the inherent inability of users to comprehend or observe the internal decision-making logic of these systems, which not only undermines trust but also raises concerns regarding the accuracy and reliability of their outputs [35,40,45,59]. Ethical concerns extend beyond technical opacity to include lack of accountability [39,58], ambiguities in the copyright of AI-generated content [48], and algorithmic biases stemming from non-transparent training data [42,57]. Particularly in the field of medical education, the lack of explainability in AI decision-making processes raises critical questions about how these systems reach conclusions that may impact student assessment, curriculum design, and ultimately patient care [35,45,57,59].The “black box” nature of these algorithms not only exacerbates distrust in AI systems [20,40,44,59] but also creates significant challenges for validating educational outcomes and ensuring ethical implementation across diverse learning environments [44,47,52,59,62,68]. This comprehensive lack of transparency necessitates the development of more explainable AI approaches and underscores the urgent need for ethical guidelines specifically addressing algorithmic transparency in medical education contexts [27,41,44,45].
Fairness.
The application of AI and LLM in medical education has drawn extensive attention to issues concerning fairness, as there are concerns that AI technology might intensify unequal resource allocation [59], particularly between high-income countries (HICs) and low-income and middle-income countries (LMICs) [26]. In the realm of medical education, the uneven distribution of medical resources, economic imbalances, technological divides, and variations in teaching standards could aggravate educational inequity [27,32,43,44,46,51,52,54,59,64,46,69]. Furthermore, the inherent bias in AI and LLM may also exacerbate unfairness towards specific groups and impact the quality of medical education [28]. For instance, due to its incapacity to consider cultural diversity in its algorithms, students from diverse cultural backgrounds may undergo unfair education [30,50]. One study has also discovered that the application of AI in ophthalmology education might be influenced by data bias, resulting in unfair services for certain groups of people [32].
Bias.
The reliance upon non-representative training data and inherent biases in data collection can give rise to discriminatory decisions against specific patient groups [24,39,48,59,61]. For example, [67] disclose that biases in surgical training data might lead to imprecise skill evaluations, while Vahedifard [70] emphasizes disparities in mental health services attributed to biased AI algorithms [44]. These issues highlight the risks of unequal treatment and diagnosis that originate from biased datasets and opaque systems. Additionally, Walsh, Stogiannos [57] have indicated that the deficiency of transparency and bias in AI within medical imaging can impact the accuracy and fairness of diagnoses.
Responsibility and accountability.
Assigning responsibility for errors in AI-driven medical decisions is a major ethical concern. The legal and moral responsibilities of virtual agents like ChatGPT in healthcare remain ambiguous [20,26,46,47,52,58,64].In the context of medical education, presents challenges across different dimensions: on one hand responsibility allocation might encompass issues like plagiarism and academic misconduct, on the other hand stemming from emerging uncertainties in technical guidance and application [27]. The research conducted by [67] suggests that the concept of responsibility may become blurred when AI technology is utilized for assessing surgical skills. Specifically, where AI systems provide erroneous assessments or misleading guidance, this tends to blur the demarcation among liable parties.
Reliability and dependence.
The questionable reliability of AI itself and excessive reliance on AI tools may weaken students’ critical thinking and problem-solving skill [39,51,55]. Dependence on AI-generated answers might impede students’ ability to independently handle complex medical issues [63]. Relying on AI tools also increases the risks of academic misconduct, such as plagiarism [53,64,68]. Research has indicated that the “hallucinations” of AI and LLM, in which the system generates or provides false or inaccurate information or data in the absence of appropriate information and may present fabricated references due to the inability to provide specific sources, have the potential to mislead medical professionals and medical students in medical education and clinical decision-making [40]. The constraints of current AI models and the quality of data also prompt people to query the reliability of the decisions they generate.For example, AI models trained on biased or outdated data may propagate distorted or erroneous cultural perspectives and clinical guidelines, particularly when applied to medical contexts beyond their original training scope. Furthermore, in complex clinical cases, analyses and diagnoses generated by AI and LLMs may lack comprehensiveness, potentially overlooking critical nuances [58,67].
Patient’s autonomy and consent.
With the incorporation of AI in medical education and healthcare, patients’ autonomy is confronted with predicaments. Regarding the collection of a large amount of patients’ data, the collection and management of such data might encroach upon patients’ autonomy [45,57]. The use of AI technology in medical education and healthcare may lead medical students to face a dilemma between relying solely on AI diagnostic recommendations and considering the patient’s symptoms and specific circumstances. Whether the patient has an exhaustive understanding of AI and whether they are informed and consent to the decision directly impacts the patient’s right to make an informed decision regarding their medical needs [45,57].
Other ethical challenges.
In addition, academic integrity and plagiarism issues, technology misuse and manipulation issues, cultural sensitivity and diversity issues, potential for malicious use, and technology-induced employment issues are also ethical challenges that need to be considered [13,20,27,31,52,53,60–62].
Risk mitigation strategies for ethical challenges
Implementing privacy protection measures.
Concerning privacy protection and data security, it is crucial to consider the issues of data privacy and security during the design and implementation of AI and LLM systems employed in medical education and embrace corresponding protective measures [20,33,35,42,46,67]. It is essential to establish agreements with relevant institutions to clarify, justify, implement, and oversee the necessity of both proactive and passive data collection, and to enhance the ethical review requirements of institutional review boards (IRBs) and national institutions to ensure their accommodation to the advancement of the AI data collection process [45]. Regulatory measures and statutes should be formulated and executed to supervise the authenticity of data and data abuse conducts, such as the EU’s GDPR prohibiting automated decision-making and data processing [60]. It is crucial to safeguard the privacy of patients and students during data processing through data encryption, data anonymization, the implementation of secure storage protocols, and the enforcement of vigorous access control [50]. For instance, in medical education settings, educators should aim to minimize the use of actual patient data in AI and LLMs and prefer to use synthetic or hypothetical scenarios to avert any potential privacy breaches. In cases in which it is necessary to input patient-specific information into AI and LLMs, it is imperative to obtain informed consent from the patient, obtain the necessary approval from an ethics committee, and implement appropriate data anonymization measures. Additionally, it is vital to provide medical students with comprehensive education and training regarding patient privacy protection, including lawful and ethical procedures for the collection, use, and storage of patient information. This foundation is crucial for fostering robust privacy awareness as AI and LLMs become increasingly pervasive in medical education. [20,33,35,42,46,67]. In the long term, AI ethics education can be incorporated in medical education to ensure that future healthcare professionals not only apprehend the ethical dimensions of AI in medicine but also are competent to make ethical decisions in practice [59].
Promoting transparency.
When developing and deploying AI models, strict adherence to the principle of transparency is indispensable, facilitating the comprehensibility and verifiability of the decision-making process and outcomes for users [22,46]. One study in2024 presented a detailed AITEL report checklist, stipulating detailed information concerning algorithm design, training data, testing/validation, etc. when reporting on AI systems to enhance transparency. Collaboration among computer scientists, medical practitioners, and ethicists should be encouraged to address the issue of algorithm opacity through multi-disciplinary cooperation. The incorporation of AI ethics education in medical education is requisite to ensure that future healthcare professionals can comprehend and handle such ethical matters [25,39,61,62].
Ensuring fairness through policy and cooperation.
To safeguard the equity of medical education in various regions and cultural milieus, international cooperation and policy support can be harnessed to aid regions endowed with limited resources in acquiring and applying AI technology. Policymakers should prioritize resource allocation and control data-sharing protocols to ensure equitable access to AI technologies [22,27,56,57]. A study suggested that through the provision of fundamental technologies and the reinforcement of AI education, it could be guaranteed that the utilization of AI technology in LMICs could satisfy local requirements [26]. When developing and utilizing AI applications, it is essential to create algorithms to identify and eliminate potential unfair factors, and adapt to different cultural backgrounds to ensure the targeted and sensitive use of AI tools [20].
Addressing bias.
Diverse datasets and algorithms should be employed in the training of AI algorithms, guaranteeing the representativeness and comprehensiveness of the data to uphold transparency for diverse patient groups and alleviate bias [28,59,67]. Regular bias audits should be carried out to identify and rectify biases present in AI systems [28].
Defining responsibility and accountability in AI and LLMs.
Specify the delimitation of responsibility for AI and LLM systems, especially in predictive and decision-making functions [45,54]. Enact laws and regulations to clarify liability for AI-related errors [41]. Instructing medical students on responsibility and accountability, while implementing rigorous evaluation and oversight mechanisms to oversee the use of AI, especially in cases where incorrect predictions lead to patient harm [22,28,32]. Highlight the role of instructors in guiding students on the proper and ethical application of AI tools.
Developing a hybrid learning model and AI application guidelines.
Adopt Embrace hybrid educational paradigms integrating AI tools with traditional teaching methodologies, stimulating students’ independent thinking and problem-solving capabilities [27,28,60]. Formulate guidelines for applying AI and LLMs in medical education and educate students about the constraints of AI to alleviate excessive reliance [32,33,56–58,62,66]. Elevate data quality, diversify data collections, and augment model transparency and interpretability to tackle AI “hallucinations” and unreliable outputs [63]. Stress human supervision and feedback mechanisms in the designs of AI systems to guarantee accuracy and reliability.
Respecting patient autonomy and data collection principles.
To implement the principle of data minimization and enhance patient autonomy in medical decision-making, we must collect only the essential minimum data required to achieve medical objectives, thereby reinforcing data security throughout the collection process [22,23,37,49]. Before data collection, it is crucial to elucidate the scope and purpose of data gathering to patients, ensuring their informed consent and understanding [22]. Additionally, enhancing the transparency and interpretability of AI algorithms will facilitate patients’ comprehension of the AI system’s decision-making process, which is vital for maintaining trust and autonomy [13]. Concurrently, intensifying education on respecting patient autonomy within medical education will equip future healthcare professionals with the ethical considerations necessary to balance technological advancements with patient rights [13,36].
Other risk mitigation strategies.
Besides the six suggestions stated above, structured training of AI in medical education, critical use of AI technologies, comprehensive reporting of AI system datasets, and explicit policies on academic misconduct can potentially mitigate or address the ethical challenges encountered when integrating AI and LLM in medical education [34,36,42,46].
Discussion
Summary of evidence
Our analysis of 50 articles reveals that the ethical landscape of AI/LLM integration in medical education is not merely a list of discrete issues but a synergistic system of tensions and trade-offs, encompassing key challenges such as privacy and data security, transparency, fairness, bias, responsibility and accountability, reliability and dependence, and patient’s autonomy and consent.
The most quantified challenge, data privacy and security (Fig 3), transcends its frequency to act as a core foundational barrier [71]. The primary ethical conflict here is not just about preventing breaches, but the philosophical dilemma of submitting sensitive, identifiable patient data into proprietary systems whose data usage protocols remain opaque [72,73]. This creates a fundamental value conflict with algorithmic transparency, as stringent privacy protections inevitably hinder external auditing for model bias and fairness [74], thereby challenging clinical acceptance and ethical deployment. For example, Rajkomar et al. argue that without access to training data and model weights, it is nearly impossible to fully diagnose and mitigate discriminatory outputs [75]. The ethical tensions between privacy and transparency directly intensify the third core challenge: fairness in global medical education. The opacity of AI systems impedes the detection of data and algorithmic biases, which can exacerbate existing global disparities and marginalize underrepresented student populations [76,77]. Beyond these core tensions, our review also identified additional intertwined challenges, including questions of responsibility and accountability, the reliability of AI outputs and the risk of over-dependence, and threats to patient autonomy and consent in data collection practices. These issues do not exist in isolation but interact with and compound the primary challenges of privacy, transparency, and fairness, reinforcing the notion of a complex ethical ecosystem.
Preliminary ethical comparison between AI and LLMs in medical education
In this study, we conducted a preliminary comparative analysis of the ethical challenges associated with the use of artificial intelligence (AI) and large language models (LLMs) in medical education. Our findings indicate that while both technologies face overarching ethical concerns—such as privacy and data security, transparency, bias, accountability, fairness, reliability, dependence, and patient autonomy—their manifestations and degrees of emphasis differ.
Notably, traditional AI applications frequently handle sensitive and identifiable patient data, which leads to more pronounced challenges pertaining to privacy and data security [78]. Furthermore, such applications often raise greater concerns regarding fairness, as they are commonly associated with biased algorithmic outcomes in assessment and resource allocation [79]. In contrast, LLMs, due to their generative nature and potential for “hallucinations”, attract more attention concerning reliability and over-dependence [68]. Additionally, owing to the opacity inherent in their automated content generation processes, LLMs also present significant challenges in transparency and accountability [80].
It should be noted that these distinctions are preliminary and based on emerging trends within the included literature. Further empirical and qualitative studies are needed to validate and refine these observations. Nonetheless, this comparative perspective offers an initial framework for understanding and addressing technology-specific ethical dilemmas in medical education, thereby aiding educators, developers, and policymakers in adopting more nuanced and effective governance strategies.
Strengths and limitations
The strengths of this scoping review are diverse and highlight its essential contribution to the field. First, a systematic search strategy was utilized across PubMed, Embase, and Web of Science, ensuring a comprehensive literature capture and enhancing the review’s reliability. Second, including perspectives from various countries reflects a global dialogue on the ethical challenges posed by AI and large language models (LLMs) in medical education. Third, the review’s timely focus on the increase in scholarly outputs from 2021 to 2024 emphasizes the relevance of these issues in the continuously evolving landscape of medical education. Fourth, many articles’ structured analysis of ethical challenges provides a clear overview of the considerations involved in integrating AI and LLMs. Lastly, the review presents concrete solutions and recommendations, offering actionable guidance for future research and policy formulation regarding the ethical application of AI in medical education. These strengths establish the review as a solid foundation for advancing the ethical discourse and practical application of AI in this field.
After reviewing the existing literature, we identified several potential limitations that may affect our understanding of AI and LLMs’ application and ethical issues in medical education. Firstly, 96.08% of the articles were in English, which could introduce bias in our perspectives, particularly regarding insights from non-English-speaking regions. Secondly, our inclusion criteria may have unintentionally excluded relevant resources, as they required articles to explicitly discuss the application of AI and LLMs in medical education or related ethical issues. Additionally, including various studies (e.g., reviews, articles, perspectives) could lead to inconsistencies in evidence quality and rigor, impacting the integration and interpretation of research findings. Lastly, while we aim for methodological rigor, selecting articles based solely on their “ethical challenges” discussion may still result in selection bias. Therefore, we must interpret these findings cautiously, recognizing the potential impact of these limitations on our results.
Implications
The findings of this review hold significant implications for future research and policy development concerning the responsible utilization of AI and LLMs in medical education. This study’s principal importance underscores the necessity of establishing ethical frameworks and guidelines for the application of AI within this domain. Such measures are essential to ensure that the integration of these technologies is both practical and aligned with ethical standards.
To address the complex ethical landscape identified in the current study, several actions are recommended:
- (1). Developing differentiated ethical guidelines: Policymakers and institutions should prioritize the creation of context-specific guidelines that differentiate between the ethical risks of traditional AI tools and those inherent to generative LLMs [81]. For instance, guidelines for LLMs must specifically address accountability for “hallucinations” and require transparency in training data sources, whereas policies for adaptive learning platforms might focus more on data sovereignty and algorithmic fairness in assessment [82]. Furthermore, there is an urgent need for clarifying legal liability [20] through legislative action, as the current ambiguity stifles innovation and adoption [83].
- (2). Integrating ethics into curriculum and faculty development: For medical educators and institutions, our synthesis implies that ethical competency must become a core component of digital health literacy. Curricula should be updated to include critical appraisal of AI-generated content, understanding of data privacy principles, and awareness of algorithmic bias [20]. The proposed “hybrid learning model” should be operationalized to position AI as a critical tool for augmentation rather than replacement, ensuring that human oversight and professional judgment remain central to clinical reasoning [84]. Institutions investing in AI tools must couple this investment with comprehensive faculty development programs focused on the ethical pedagogical use of these technologies.
- (3). Focusing future research on critical gaps: This review highlights critical gaps that demand targeted future inquiry. Future research should employ qualitative methods (e.g., deliberative workshops) to explore the ethical trade-offs stakeholders are willing to make (e.g., trading some privacy for greater educational efficacy) and to validate the preliminary differentiation between AI and LLM ethics proposed here [85]. Furthermore, rather than merely describing the problem, interventional studies are needed to develop and test models for equitable AI access and implementation in low-resource settings, moving from theory to action. The complex interplay of challenges requires deep interdisciplinary collaboration; future work should co-create solutions with teams comprising ethicists, clinicians, data scientists, lawyers, and patients to ensure they are robust, practical, and ethically sound. Lastly, research must move beyond conceptual analysis to longitudinal studies that track the real-world ethical outcomes of AI integration in medical education over time.
This review provides a foundational framework for future research and policy-making, fostering advancements in medical education practices, promoting global health equity, considering long-term ethical considerations, and enhancing interdisciplinary collaboration. We can strive toward a more moral and practical integration of AI and LLMs in medical education by addressing the identified challenges and adhering to the proposed measures and research directions.
Conclusion
Medical ethics principles are fundamentally constituted by four core tenets: respect for autonomy, beneficence, non-maleficence, and justiceThe integration of AI and LLMs into medical education presents a dualistic landscape of immense potential and significant ethical complexity. Our scoping review identifies a core tension: these technologies risk undermining the very ethical principles they aim to teach, particularly respect for autonomy (through privacy violations and opaque decision-making) and justice (through perpetuated algorithmic biases). The defining mission of medical education(to cultivate knowledgeable, skilled, and ethically-grounded practitioners) demands that this integration be guided not solely by technological capability but by a steadfast commitment to medical ethics. Therefore, AI and LLMs should not be viewed merely as tools for efficiency, but as catalysts for developing deeper ethical reasoning and digital literacy among future healthcare professionals. Moving forward, the field must transition from conceptual mapping to actionable solutions. This necessitates robust empirical research to evaluate the real-world impact of AI on learning and ethical development, a dedicated focus on ensuring these technologies promote rather than hinder global health equity, and a critical examination of their long-term implications for healthcare. Ultimately, the goal is a future where AI integration strengthens, rather than compromises, the foundation of ethical medical practice.
Supporting information
S2 File. The specific search terms for each database.
https://doi.org/10.1371/journal.pone.0333411.s002
(DOCX)
S1 Table. Summary of Studies Identified and Screened in the Literature Search.
https://doi.org/10.1371/journal.pone.0333411.s004
(XLSX)
S2 Table. The standard characteristics of study table.
https://doi.org/10.1371/journal.pone.0333411.s005
(XLSX)
Acknowledgments
Statement on the Use of AI Tools: We acknowledge the assistance provided by Kimi AI in the process of data collection and organization. All content generated by Kimi AI was thoroughly reviewed and validated by the research team to ensure its accuracy and reliability. The research team takes full responsibility for the accuracy and integrity of the final manuscript.
References
- 1. Kolanska K, Chabbert-Buffet N, Daraï E, Antoine J-M. Artificial intelligence in medicine: A matter of joy or concern? J Gynecol Obstet Hum Reprod. 2021;50(1):101962. pmid:33148398
- 2. Ghassemi M, Birhane A, Bilal M, Kankaria S, Malone C, Mollick E, et al. ChatGPT one year on: who is using it, how and why? Nature. 2023;624(7990):39–41. pmid:38036860
- 3. Thirunavukarasu AJ, Ting DSJ, Elangovan K, Gutierrez L, Tan TF, Ting DSW. Large language models in medicine. Nat Med. 2023;29(8):1930–40. pmid:37460753
- 4. Sallam M. ChatGPT Utility in Healthcare Education, Research, and Practice: Systematic Review on the Promising Perspectives and Valid Concerns. Healthcare (Basel). 2023;11(6):887. pmid:36981544
- 5. Abdullahi T, Singh R, Eickhoff C. Learning to Make Rare and Complex Diagnoses With Generative AI Assistance: Qualitative Study of Popular Large Language Models. JMIR Med Educ. 2024;10:e51391. pmid:38349725
- 6. Narayanan S, Ramakrishnan R, Durairaj E, Das A. Artificial Intelligence Revolutionizing the Field of Medical Education. Cureus. 2023;15(11):e49604. pmid:38161821
- 7. Schwarz MR, Wojtczak A. Global minimum essential requirements: a road towards competence-oriented medical education. Med Teach. 2002;24(2):125–9. pmid:12098430
- 8. Hersh W, Fultz Hollis K. Results and implications for generative AI in a large introductory biomedical and health informatics course. NPJ Digit Med. 2024;7(1):247. pmid:39271955
- 9. Safranek CW, Sidamon-Eristoff AE, Gilson A, Chartash D. The Role of Large Language Models in Medical Education: Applications and Implications. JMIR Med Educ. 2023;9:e50945. pmid:37578830
- 10. Benítez TM, Xu Y, Boudreau JD, Kow AWC, Bello F, Van Phuoc L, et al. Harnessing the potential of large language models in medical education: promise and pitfalls. J Am Med Inform Assoc. 2024;31(3):776–83. pmid:38269644
- 11. Abd-Alrazaq A, AlSaad R, Alhuwail D, Ahmed A, Healy PM, Latifi S, et al. Large Language Models in Medical Education: Opportunities, Challenges, and Future Directions. JMIR Med Educ. 2023;9:e48291. pmid:37261894
- 12. Pasek J. The virtues in medical practice. J Med Ethics. 1995;21(3):188.1–188.
- 13. Lu H, Alhaskawi A, Dong Y, Zou X, Zhou H, Ezzi SHA, et al. Patient Autonomy in Medical Education: Navigating Ethical Challenges in the Age of Artificial Intelligence. Inquiry. 2024;61:469580241266364. pmid:39290068
- 14. Harishbhai Tilala M, Kumar Chenchala P, Choppadandi A, Kaur J, Naguri S, Saoji R, et al. Ethical Considerations in the Use of Artificial Intelligence and Machine Learning in Health Care: A Comprehensive Review. Cureus. 2024;16(6):e62443. pmid:39011215
- 15. Radanliev P, Santos O, Brandon-Jones A, Joinson A. Ethics and responsible AI deployment. Front Artif Intell. 2024;7:1377011. pmid:38601110
- 16. Tricco AC, Lillie E, Zarin W, O’Brien KK, Colquhoun H, Levac D, et al. PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Ann Intern Med. 2018;169(7):467–73. pmid:30178033
- 17.
A I M. An AI assistant that can perform deductive reasoning and is capable of deep thinking. Available from: https://kimi.moonshot.cn
- 18.
Group NM. Kimi AI’s breakthroughs in 2025: What you need to know. Available from: https://www.nitromediagroup.com/kimi-ai-2025-breakthroughs-redefining-ai/. 2025.
- 19.
Arksey H, O’Malley L. Scoping studies: Framework, Towards a methodological. 2005.
- 20. Alam F, Lim MA, Zulkipli IN. Integrating AI in medical education: embracing ethical usage and critical understanding. Front Med (Lausanne). 2023;10:1279707. pmid:37901398
- 21. Borger JG, Ng AP, Anderton H, Ashdown GW, Auld M, Blewitt ME, et al. Artificial intelligence takes center stage: exploring the capabilities and implications of ChatGPT and other AI-assisted technologies in scientific research and education. Immunol Cell Biol. 2023;101(10):923–35. pmid:37721869
- 22. Busch F, Adams LC, Bressem KK. Biomedical Ethical Aspects Towards the Implementation of Artificial Intelligence in Medical Education. Med Sci Educ. 2023;33(4):1007–12. pmid:37546190
- 23. Çalışkan SA, Demir K, Karaca O. Artificial intelligence in medical education curriculum: An e-Delphi study for competencies. PLoS One. 2022;17(7):e0271872. pmid:35862401
- 24. Civaner MM, Uncu Y, Bulut F, Chalil EG, Tatli A. Artificial intelligence in medical education: a cross-sectional needs assessment. BMC Med Educ. 2022;22(1):772. pmid:36352431
- 25. Cornwall J, Hildebrandt S, Champney TH, Goodman K. Ethical concerns surrounding artificial intelligence in anatomy education: Should AI human body simulations replace donors in the dissection room? Anat Sci Educ. 2024;17(5):937–43. pmid:37750493
- 26. Ejaz H, McGrath H, Wong BL, Guise A, Vercauteren T, Shapey J. Artificial intelligence and medical education: A global mixed-methods study of medical students’ perspectives. Digit Health. 2022;8:20552076221089099. pmid:35521511
- 27. Fatima SS, Sheikh NA, Osama A. Authentic assessment in medical education: exploring AI integration and student-as-partners collaboration. Postgrad Med J. 2024;100(1190):959–67. pmid:39041454
- 28. Franco D’Souza R, Mathew M, Mishra V, Surapaneni KM. Twelve tips for addressing ethical concerns in the implementation of artificial intelligence in medical education. Med Educ Online. 2024;29(1):2330250. pmid:38566608
- 29. Ghaempanah F, Moasses Ghafari B, Hesami D, Hossein Zadeh R, Noroozpoor R, Moodi Ghalibaf A, et al. Metaverse and its impact on medical education and health care system: A narrative review. Health Sci Rep. 2024;7(9):e70100. pmid:39323461
- 30. Ghorashi N, Ismail A, Ghosh P, Sidawy A, Javan R. AI-Powered Chatbots in Medical Education: Potential Applications and Implications. Cureus. 2023;15(8):e43271. pmid:37692629
- 31. Grunhut J, Marques O, Wyatt ATM. Needs, Challenges, and Applications of Artificial Intelligence in Medical Education Curriculum. JMIR Med Educ. 2022;8(2):e35587. pmid:35671077
- 32. Gurnani B, Kaur K. Leveraging ChatGPT for ophthalmic education: A critical appraisal. Eur J Ophthalmol. 2024;34(2):323–7. pmid:37974429
- 33. Hong H, Kang Y, Kim Y, Kim B. Application of artificial intelligence in medical education: focus on the application of ChatGPT for clinical medical education. J Med Life Sci. 2023;20(2):53–9.
- 34. Jackson P, Ponath Sukumaran G, Babu C, Tony MC, Jack DS, Reshma VR, et al. Artificial intelligence in medical education - perception among medical students. BMC Med Educ. 2024;24(1):804. pmid:39068482
- 35. Kimmerle J, Timm J, Festl-Wietek T, Cress U, Herrmann-Werner A. Medical Students’ Attitudes toward AI in Medicine and their Expectations for Medical Education. Cold Spring Harbor Laboratory. 2023.
- 36. Knopp MI, Warm EJ, Weber D, Kelleher M, Kinnear B, Schumacher DJ, et al. AI-Enabled Medical Education: Threads of Change, Promising Futures, and Risky Realities Across Four Potential Future Worlds. JMIR Med Educ. 2023;9:e50373. pmid:38145471
- 37. Koçer Tulgar Y, Tulgar S, Güven Köse S, Köse HC, Çevik Nasırlıer G, Doğan M, et al. Anesthesiologists’ Perspective on the Use of Artificial Intelligence in Ultrasound-Guided Regional Anaesthesia in Terms of Medical Ethics and Medical Education: A Survey Study. Eurasian J Med. 2023;55(2):146–51. pmid:37161553
- 38. Rodriguez Peñaranda N, Eissa A, Ferretti S, Bianchi G, Di Bari S, Farinha R, et al. Artificial Intelligence in Surgical Training for Kidney Cancer: A Systematic Review of the Literature. Diagnostics (Basel). 2023;13(19):3070. pmid:37835812
- 39. Lee H. The rise of ChatGPT: Exploring its potential in medical education. Anat Sci Educ. 2024;17(5):926–31. pmid:36916887
- 40.
Lee H. Reflections on the development of large language models and their impact on basic medical education. 2024. https://doi.org/10.11637/aba.2024.37.2.59
- 41. Lee J, Wu AS, Li D, Kulasegaram KM. Artificial Intelligence in Undergraduate Medical Education: A Scoping Review. Acad Med. 2021;96(11S):S62–70. pmid:34348374
- 42. Zhui L, Yhap N, Liping L, Zhengjie W, Zhonghao X, Xiaoshu Y, et al. Impact of Large Language Models on Medical Education and Teaching Adaptations. JMIR Med Inform. 2024;12:e55933. pmid:39087590
- 43. Lucas HC, Upperman JS, Robinson JR. A systematic review of large language models and their implications in medical education. Med Educ. 2024;58(11):1276–85. pmid:38639098
- 44. Ma M, Li Y, Gao L, Xie Y, Zhang Y, Wang Y, et al. The need for digital health education among next-generation health workers in China: a cross-sectional survey on digital health education. BMC Med Educ. 2023;23(1):541. pmid:37525126
- 45. Masters K. Ethical use of Artificial Intelligence in Health Professions Education: AMEE Guide No. 158. Med Teach. 2023;45(6):574–84. pmid:36912253
- 46. Masters K, Salcedo D. A checklist for reporting, reading and evaluating Artificial Intelligence Technology Enhanced Learning (AITEL) research in medical education. Med Teach. 2024;46(9):1175–9. pmid:38227374
- 47. Mei Z, Jin S, Li W, Zhang S, Cheng X, Li Y, et al. Ethical risks in robot health education: A qualitative study. Nurs Ethics. 2025;32(3):913–30. pmid:39138639
- 48. Mohammad B, Supti T, Alzubaidi M, Shah H, Alam T, Shah Z. The pros and cons of using ChatGPT in medical education: a scoping review. Healthcare Transformation with Informatics and Artificial Intelligence. 2023.
- 49. Perchik JD, Smith AD, Elkassem AA, Park JM, Rothenberg SA, Tanwar M, et al. Artificial Intelligence Literacy: Developing a Multi-institutional Infrastructure for AI Education. Acad Radiol. 2023;30(7):1472–80. pmid:36323613
- 50. Salih SM. Perceptions of Faculty and Students About Use of Artificial Intelligence in Medical Education: A Qualitative Study. Cureus. 2024;16(4):e57605. pmid:38707183
- 51. Shimizu I, Kasai H, Shikino K, Araki N, Takahashi Z, Onodera M, et al. Developing Medical Education Curriculum Reform Strategies to Address the Impact of Generative AI: Qualitative Study. JMIR Med Educ. 2023;9:e53466. pmid:38032695
- 52. Silver JK, Dodurgali MR, Gavini N. Artificial Intelligence in Medical Education and Mentoring in Rehabilitation Medicine. Am J Phys Med Rehabil. 2024;103(11):1039–44. pmid:39016292
- 53. Skryd A, Lawrence K. ChatGPT as a Tool for Medical Education and Clinical Decision-Making on the Wards: Case Study. JMIR Form Res. 2024;8:e51346. pmid:38717811
- 54. Sun L, Yin C, Xu Q, Zhao W. Artificial intelligence for healthcare and medical education: a systematic review. Am J Transl Res. 2023;15(7):4820–8. pmid:37560249
- 55. Turner L, Hashimoto DA, Vasisht S, Schaye V. Demystifying AI: Current State and Future Role in Medical Education Assessment. Acad Med. 2024;99(4S Suppl 1):S42–7. pmid:38166201
- 56. Vahedifard F, Haghighi AS, Dave T, Tolouei M, Zare FH. Practical Use of ChatGPT in Psychiatry for Treatment Plan and Psychoeducation. Arxiv. 2023.
- 57. Walsh G, Stogiannos N, van de Venter R, Rainey C, Tam W, McFadden S, et al. Responsible AI practice and AI education are central to AI implementation: a rapid review for all medical imaging professionals in Europe. BJR Open. 2023;5(1):20230033. pmid:37953871
- 58. Wang H, Wu W, Dou Z, He L, Yang L. Performance and exploration of ChatGPT in medical examination, records and education in Chinese: Pave the way for medical AI. Int J Med Inform. 2023;177:105173. pmid:37549499
- 59. Weidener L, Fischer M. Teaching AI Ethics in Medical Education: A Scoping Review of Current Literature and Practices. Perspect Med Educ. 2023;12(1):399–410. pmid:37868075
- 60. Weidener L, Fischer M. Artificial Intelligence in Medicine: Cross-Sectional Study Among Medical Students on Application, Education, and Ethical Aspects. JMIR Med Educ. 2024;10:e51247. pmid:38180787
- 61. Weidener L, Fischer M. Proposing a Principle-Based Approach for Teaching AI Ethics in Medical Education. JMIR Med Educ. 2024;10:e55368. pmid:38285931
- 62. Wong RS-Y, Ming LC, Raja Ali RA. The Intersection of ChatGPT, Clinical Medicine, and Medical Education. JMIR Med Educ. 2023;9:e47274. pmid:37988149
- 63. Wu Y, Zheng Y, Feng B, Yang Y, Kang K, Zhao A. Embracing ChatGPT for Medical Education: Exploring Its Impact on Doctors and Medical Students. JMIR Med Educ. 2024;10:e52483. pmid:38598263
- 64. Xu T, Weng H, Liu F, Yang L, Luo Y, Ding Z, et al. Current Status of ChatGPT Use in Medical Education: Potentials, Challenges, and Strategies. J Med Internet Res. 2024;26:e57896. pmid:39196640
- 65. Zarei M, Eftekhari Mamaghani H, Abbasi A, Hosseini M-S. Application of artificial intelligence in medical education: A review of benefits, challenges, and solutions. Medicina Clínica Práctica. 2024;7(2):100422.
- 66. Zarei M, Zarei M, Hamzehzadeh S, Shakeri Bavil Oliyaei S, Hosseini M-S. ChatGPT, a Friend or a Foe in Medical Education: A Review of Strengths, Challenges, and Opportunities. Shiraz E-Med J. 2024;25(6).
- 67. Rodriguez Peñaranda N, Eissa A, Ferretti S, Bianchi G, Di Bari S, Farinha R, et al. Artificial Intelligence in Surgical Training for Kidney Cancer: A Systematic Review of the Literature. Diagnostics (Basel). 2023;13(19):3070. pmid:37835812
- 68. Zhui L, Fenghe L, Xuehu W, Qining F, Wei R. Ethical Considerations and Fundamental Principles of Large Language Models in Medical Education: Viewpoint. J Med Internet Res. 2024;26:e60083. pmid:38971715
- 69. Sun L, Yin C, Xu Q, Zhao W. Artificial intelligence for healthcare and medical education: a systematic review. Am J Transl Res. 2023;15(7):4820–8. pmid:37560249
- 70.
Vahedifard F. Practical use of ChatGPT in psychiatry for treatment plan and psychoeducation. 2023.
- 71. Price WN 2nd, Cohen IG. Privacy in the age of medical big data. Nat Med. 2019;25(1):37–43. pmid:30617331
- 72.
Mccradden MD, Joshi S, Mazwi M, Anderson J. Ethical limitations of algorithmic fairness solutions in health care machine learning. 2020;2(5).
- 73. Char DS, Shah NH, Magnus DJ. Implementing machine learning in health care - addressing ethical challenges. N Engl J Med. 2018;378(11):981–3.
- 74.
Marey A, Arjmand P, Alerab ADS, Eslami MJ, Saad AM, Sanchez N. Explainability, transparency and black box challenges of AI in radiology: impact on patient care in cardiovascular radiology. 2024;55(1).
- 75. Rajkomar A, Hardt M, Howell MD, Corrado G, Chin MH. Ensuring fairness in machine learning to advance health equity. Am J Ind Med. 2018.
- 76. Chen RJ, Wang JJ, Williamson DFK, Chen TY, Lipkova J, Lu MY, et al. Algorithmic fairness in artificial intelligence for medicine and healthcare. Nat Biomed Eng. 2023;7(6):719–42. pmid:37380750
- 77. Grote T, Keeling G. On Algorithmic Fairness in Medical Practice. Camb Q Healthc Ethics. 2022;31(1):83–94. pmid:35049447
- 78. Stogiannos N, Georgiadou E, Rarri N, Malamateniou C. Ethical AI: A qualitative study exploring ethical challenges and solutions on the use of AI in medical imaging. European Journal of Radiology Artificial Intelligence. 2025;1:100006.
- 79. Lu K, Sun S, Liu W, Jiang J, Yan Z. Mapping Key Nodes and Global Trends in AI and Large Language Models for Medical Education: A Bibliometric Study. Adv Med Educ Pract. 2025;16:1421–38. pmid:40832621
- 80.
de Cerqueira JS, Kemell KK, Rousi R, Xi N, Hamari J, Abrahamsson P. Mapping trustworthiness in large language models: A bibliometric analysis bridging theory to practice. 2025.
- 81.
Radanliev PJAAI. AI ethics: Integrating transparency, fairness, and privacy in AI development. 2025;39(1).
- 82.
Organization WH. Ethics and governance of artificial intelligence for health: Guidance on large multi-modal models. Geneva: World Health Organization. 2024. https://www.who.int/publications/i/item/9789240084759
- 83.
Ferlito B, Segers S, Proost MD, Mertes HJS, ethics e. Responsibility gap(s) due to the introduction of AI in healthcare: an ubuntu-inspired approach. 30(4):34.
- 84.
Lu H, Ahmad Alhaskawi M, Yanzhao Dong M, Xiaodi Zou M, Haiying Zhou M, Mmed SHAE. Patient autonomy in medical education: navigating ethical challenges in the age of artificial intelligence. 2024;61(000).
- 85. Schneider D, Liedtke W, Klausen AD, Lipprandt M, Funer F, Bratan T, et al. Indecision on the use of artificial intelligence in healthcare-A qualitative study of patient perspectives on trust, responsibility and self-determination using AI-CDSS. Digit Health. 2025;11:20552076251339522. pmid:40469779