Given the increasing importance and potential impact of artificial intelligence (AI) technology on healthcare delivery, there is growing demand for integrating AI into medical school curricula [1]. Cooper and Rodman [2] emphasized that Pandora’s box of AI has already opened, urging medical schools to begin the arduous process of incorporating AI didactics into clinical skills courses, diagnostic reasoning lessons, and training in system-based practice. Lee et al. [3] highlighted that the physicians’ ability to manage AI technologies should be considered an essential competence and emphasized the necessity for AI training in the undergraduate medical education curriculum. Domrös-Zoungrana et al. [4] emphasized the importance of incorporating both ‘learning about AI’ and ‘learning with AI’ into medical education to effectively prepare future healthcare professionals.

Despite this recognized need, several challenges have been identified in integrating AI into medical education. Limited time within the already dense curricula, faculty resistance, misunderstandings of AI, lack of faculty expertise in AI, varying levels of AI literacy among faculty members, and insufficient infrastructure have been identified as significant obstacles to integrating AI into medical education [4,5]. Previous research has proposed a framework for integrating AI into the medical supported by the of existing literature to articulate the authors’ perspectives [3-5] or empirical studies [6]. Previous studies have also attempted to identify learners’ needs regarding AI education in medical curricula. For example, a cross-sectional, multi-center study conducted in Turkey with 3,018 medical students nationwide revealed a strong demand for updates to the medical curriculum to address the evolving needs of healthcare transformation driven by AI [7].

Practical guidelines and strategies for implementing a medical AI framework are essential to develop the AI competencies required by both current and future physicians, while also aligning with existing medical curricula. Additionally, when integrating a new domain into an existing curriculum, it is crucial to conduct a needs analysis that reflects stakeholders’ voices within their specific contexts. This study aimed to gather the perceptions of faculty members and students regarding the integration of AI education into medical curricula in the Korean context.

Three key themes emerged regarding faculty and student perceptions of integrating AI education into medical curricula: (1) learning content, (2) curricular design and instructional methods, and (3) concerns regarding AI education in medical schools. Regarding the content of AI education in medical schools, the participants emphasized the importance of focusing on foundational knowledge, clinical applications, and addressing ethical and legal considerations. They also highlighted the need to better understand the potential limitations of AI in medical practice to mitigate both hype and undue skepticism (Table 1).

In terms of curricular design and instructional methods, adopting a spiral curriculum introducing basic AI concepts in the early stages of medical education and gradually progressing to more complex topics in later stages was suggested. Respondents also advocated for a tailored learning approach that included both mandatory and elective courses; diverse instructional methods, including hands-on practice; exposure to industry or expert developers; and common AI education programs across medical schools (Table 2).

Several concerns regarding AI arose. Respondents expressed apprehension about balancing AI integration within the principal goals of medical education, noting that an excessive emphasis on AI may detract from the core objective of training competent medical professionals. They also cited time constraints and the additional academic burden of incorporating AI into an already dense curriculum. The lack of qualified experts teaching medical AI has been highlighted as an important issue. Furthermore, they suggested that integrating AI education for all medical students might be premature, advocating a more cautious approach until the technology matures (Table 3).

The learning content for medical AI identified in this study aligns closely with the medical AI competencies recommended in prior research [3,4,6,7], particularly in foundational AI knowledge, clinical applications, and ethical and legal considerations. A large-scale nationwide survey on AI education needs assessment emphasized the necessity of updated education on AI applications, ethical issue management, and safeguarding professional values and rights [7]. Similarly, a scoping review of AI in undergraduate medical education underscored the need for an AI curriculum while highlighting the lack of consensus regarding its content and delivery methods [3]. In response to these findings, Lee et al. [6] developed a set of medical AI competencies for medical graduates and suggested a dual approach of mandatory and optional studies for AI education. However, their work did not address curricular design or instructional methods tailored to the needs of students and educators.

This study reveals that both faculty and students prioritize the importance of practical and clinical AI applications. Specifically, they emphasized real-world use cases and interpretation skills over technical development, reflecting the need for a hands-on, application-oriented approach to medical AI education. Integrating case-based learning grounded in real-world scenarios involving AI tools would enhance the learning of this content. The participants also emphasized the importance of viewing medical AI as an assistive tool to improve diagnostic accuracy and reduce physician workload. The respondents of this study supported the spiral integration of AI in education, which reflects a well-established approach to curriculum design. This method emphasizes helping students to progressively strengthen and deepen their understanding of content through repeated systematic engagement over time [8]. This fosters a gradual and comprehensive grasp of AI, reducing hesitancy, resistance, and misunderstandings about the risks associated with AI owing to a lack of knowledge. Stakeholder support for implementing both mandatory and elective AI courses aligns with the growing focus on tailoring education to learners’ motivations and career aspirations in contemporary curriculum design [9]. Additionally, an individualized approach offers the advantage of alleviating concerns about curriculum overload while accommodating varying levels of student interest.

Addressing the concerns surrounding AI education, particularly the need to balance AI integration with the fundamental goals of medical education amidst contemporary advancements, requires medical schools to define the goals and outcomes of AI education in alignment with the broader framework of basic medical education outcomes. The integration of AI education should be flexible and adapt to technological advancements, ensuring that it complements and enhances overall educational objectives. Additionally, addressing concerns regarding resource shortages requires careful planning and allocation of both physical and human resources. As faculty members highlighted, developing the expertise of educators is crucial to effectively deliver AI education. Medical schools should support faculty development to ensure that they are not only proficient in AI but also integrate interdisciplinary knowledge, including ethics and clinical applications, into their teaching [10].

In conclusion, although our study was conducted within the Korean context and has the limitation of not representing the opinions of all students and faculty members, it may add valuable insights into the content and methods that should be prioritized when designing AI education. Moreover, given the rapid evolution of both medical learners and AI technologies, ongoing, timely needs assessment of AI curriculum development is essential to ensure relevance and effectiveness.