The global higher education landscape is currently undergoing a period of profound transition, marked by the rapid convergence of digital technologies and pedagogical theories. Within the sphere of medical education, this transformation is particularly acute, as the maturation of artificial intelligence (AI) necessitates a comprehensive reevaluation of the competencies required of both academic faculty and their students. AI represents one of the most disruptive innovations in healthcare, attracting intense focus from physicians, researchers, and educational leaders due to its ability to manage massive quantities of unstructured data and solve complex clinical problems [1–3]. This integration is no longer a distant forecast but a present-day reality, with recent data indicating that 2 in 3 physicians are already using health AI in their clinical practice, marking a 78% increase from 2023 [4, 5].

This technological shift demands a fundamental reconceptualization of how medical knowledge is taught, assessed, and applied [6, 7]. Large language models (LLMs) such as GPT-4 have demonstrated remarkable capabilities in clinical reasoning, differential diagnosis, and data interpretation [5, 8]. Evidence shows that LLMs can perform at levels comparable to or exceeding medical students and residents on standardized medical examinations, such as the USMLE [6–9]. However, the integration of AI into medical training remains highly inconsistent [10]. A 2024 scoping review revealed a critical gap: while 71.8% of German medical schools offered AI-related courses, the vast majority remained elective or extracurricular, and 85% of Canadian medical students reported no formal AI education within their core curriculum [11, 12]. Furthermore, a Stanford Medicine survey found that 44% of physicians and 23% of medical students believed their education had not adequately prepared them for emerging healthcare technologies [13].

Ukraine’s medical education system faces additional, unique pressures that have simultaneously accelerated digital adoption and tightened resource constraints [14]. The dual impact of the COVID-19 pandemic and the full-scale military aggression since 2022 compelled a massive shift toward online and blended learning models, requiring academic faculty to attain unprecedented digital literacy in a condensed period. National strategies identify digital transformation as an operational goal essential for ensuring the international competitiveness of Ukrainian graduates. Despite these strategic aims, a significant gap exists between policy and binding regulation. To date, no official national regulations have defined the legal boundaries of AI use in academic research or clinical training in Ukraine. The «White Paper» issued by the Ministry of Digital Transformation provides general guidance but lacks enforceable rules, leaving medical universities to navigate ethical complexities independently [8]. To address this regulatory gap, Ivano-Frankivsk National Medical University formally adopted a comprehensive «Policy on the Use of Artificial Intelligence Systems» in May 2025 [15]. This policy establishes a binding legal and ethical framework for integrating AI in educational, scientific, and administrative activities, aligning with national recommendations from the Ministry of Digital Transformation and the Ministry of Education and Science of Ukraine.

The urgency of formal training is underscored by the high rate of «bottom-up» adoption among Ukrainian learners. Research at Bogomolets National Medical University indicates that 84% of medical PhD students and interns utilize AI tools for academic purposes. However, this adoption is plagued by ethical uncertainty and a lack of structured guidance: 51% of participants in the same study admitted to cheating on tests previously, and 36% of students viewed the use of AI itself as a form of misconduct [8, 16]. There is a critical danger that students may stop thinking critically or fall victim to «automation bias», accepting AI outputs without question [5]. Furthermore, the risk of AI «hallucinations» poses a direct threat to patient safety if clinical decisions are based on unverified AI outputs [3, 4, 17].

Addressing this bottleneck requires a shift in the educator’s role from a «transmitter of absolute truths» to a «mediator of learning» and knowledge curator [18, 19]. Systematic faculty training is essential for creating institutional capacity for AI-integrated curriculum development and establishing robust ethical frameworks. Academic faculty must be prepared to navigate the shift from the traditional clinician-patient dyad to a much more complex, ethically and emotionally, clinician-AI-patient triad [20]. By treating AI training as a «humanitarian category» that encompasses cultural reproduction and professional identity, rather than a purely technical skill, institutions can foster unique human abilities, such as empathy and ethical judgment, which complement machine capabilities [21].

Against this backdrop, Ivano-Frankivsk National Medical University (IFNMU) launched Ukraine’s first comprehensive, systematic program to train medical faculty in AI technologies. This initiative was designed as a foundational element of institutional digital transformation rather than an isolated intervention. The program aimed to build faculty digital literacy, create capacity for AI-integrated assignments, and demonstrate a scalable model for other institutions seeking to maintain academic rigor and ethical standards in an increasingly AI-driven healthcare environment. This article describes the development, implementation, and outcomes of this pioneering program, offering practical insights for the future of medical pedagogy in Ukraine and beyond.

The final program structure consisted of four interconnected components designed to support academic faculty through various stages of adoption. Component 1 consisted of structured, practical training workshops, each lasting 3 to 4 hours in duration. These in-person sessions covered AI fundamentals, large language models (LLMs), and specific tools for generating questions, clinical case studies, and literature reviews. Component 2 offered individual consultations, lasting 60–90 minutes, available both in person and via video conference. These sessions allowed facilitators to address discipline-specific challenges, such as developing AI-integrated assignments or designing assessment strategies that mitigate academic dishonesty.

Component 3 focused on the collaborative development of methodological materials, including a comprehensive institutional guide approved by the Academic Council, twelve discipline-specific prompt engineering guides, and eighteen assessment rubrics. Component 4 established a community of practice for ongoing peer support, utilizing dedicated communication channels and monthly meetups to share successful use cases and innovative applications. The program employed a voluntary but highly encouraged recruitment strategy, involving outreach from department heads and testimonials from early adopters to foster a culture of sustainable transformation rather than mandated compliance.

The training program integrated a diverse range of AI platforms, categorized by their primary functional application in medical education and research. For general text generation and high-quality dialogue, faculty utilized OpenAI’s ChatGPT, which reached 100 million users within two months of launch, as well as Microsoft Copilot for its integration with office productivity suites. Anthropic’s Claude was emphasized for ethically guided interactions, as its development is governed by a specific «constitution» of safety principles. For real-time information retrieval, the program included xAI Grok, noted for its integration with current data from X (formerly Twitter), and Perplexity AI, which is favored in academic settings for its ability to provide direct citations for its outputs.

For scientific research and systematic reviews, academic faculty were trained on Elicit. This AI research assistant identifies papers based on meaning rather than keywords, and SciSpace, which extracts key theses, methods, and conclusions from academic articles. The program also introduced Connected Papers to visualize citation graphs. To organize custom knowledge bases, faculty employed NotebookLM, which analyzes personal research files to provide structured summaries.

Methodological and visual support tools included Napkin AI, which generates diagrams and infographics from raw text. For creating structured presentations, the curriculum included Gamma, which enables the generation of interactive slides from textual descriptions, and Genspark, which provides automated structural content generation. Language and communication support relied on DeepL for neural machine translation and Grammarly AI for refining the grammar, style, and tone of English-language academic manuscripts.

A core methodology for «prompt engineering» was established to help faculty overcome the «blank page problem» and improve the accuracy of AI outputs. Training followed a multi-stage logic refinement process:

Step 1: Formulating a simple request.

Step 2: Adding context and a specific target audience.

Step 3: Specifying the response format.

Step 4: Adding desired characteristics, such as the use of professional medical terminology.

Step 5: Refining the style for clarity and asking the AI to provide verifiable scientific references to reduce the risk of «hallucinations».

Faculty were also instructed in the use of specialized personas, such as teaching the AI to «Adopt the role of a Professor of Biology» to ensure the AI utilizes a specialized vocabulary from its database. To stimulate creativity, educational gaming via tools like Secret Prompter was used to help faculty experiment with unique generative instructions.

The study utilized a mixed-methods approach to evaluate program outcomes. Quantitative metrics included the tracking of faculty completion rates across departments, the number of resources developed (including a repository of 142 use cases), and student satisfaction ratings from deployed AI assistants. Qualitative data collection involved semi-structured interviews with thirty-five program participants representing diverse ranks (Assistant, Associate, and Full Professors) and departments. Semi-structured interviews were conducted using an interview guide developed specifically for this study (see Additional File 1). The guide covered seven main areas: background and prior experience with digital technologies, training program experience and satisfaction, implementation and integration of AI tools into teaching practice, ethical considerations and concerns, perceived impact and value of the program, recommendations for future directions, and opportunities for additional comments. The interview guide was developed by the research team in Ukrainian (the primary language of instruction at IFNMU) and pilot-tested with three faculty members before full implementation. Interviews lasted 30–45 minutes and were conducted either in person or via video conference based on participant preference. All interviews were audio-recorded with permission and transcribed verbatim for analysis. Additionally, feedback was collected from 847 students across AI-enhanced courses using existing institutional evaluation mechanisms. Qualitative data were analyzed through a thematic analysis conducted by a team of educational researchers to identify recurring patterns in faculty experiences and challenges to implementation.

The implementation of the AI training program at Ivano-Frankivsk National Medical University addresses a critical global bottleneck in medical education: the shortage of faculty expertise in utilizing advanced digital technologies. While international scoping reviews indicate that over 70% of German medical schools offer AI-related content, these initiatives are often elective, leaving 85% of Canadian medical students, for example, without formal AI training in their core curriculum. Our results, demonstrating a 29% participation rate among the university’s total faculty (211 out of 732 scientific and pedagogical staff members), suggest that even as a voluntary initiative, systematic training can achieve significant institutional reach and overcome the typical pattern where digital adoption clusters primarily among early adopters.

This high engagement is attributed to three primary factors: a supportive institutional leadership that allocated resources for digital infrastructure, a modular curriculum scaffolded into three competency levels, and the creation of a Community of Practice that reduced professional isolation. By treating AI training as a humanitarian and philosophical category—rather than a purely technical skill—the program allowed educators to transition from a «transmitter of absolute truths» to a «mediator of learning» and curator of knowledge. This shift is essential for navigating the evolving clinician-AI-patient triad, where machine intelligence complements rather than replaces human expertise.

The diversity of tools integrated into the IFNMU program highlights the extensive applicability of AI across various medical disciplines. Beyond general language models like ChatGPT, which reached 100 million users in two months, the program utilized specialized research tools. Elicit and SciSpace were used to automate literature reviews for faculty. The introduction of Perplexity AI was particularly significant for academic research due to its ability to provide direct citations, which mitigates the risk of AI «hallucinations».

A core methodological contribution of this study is the formalization of prompt engineering within medical pedagogy. The «blank page problem» was addressed through a structured five-step logic refinement process, which moved faculty from simple requests to complex, context-rich instructions. By using specialized personas, such as instructing the AI to «Adopt the role of a Professor of Biology», educators were able to ensure that generative models utilized the appropriate professional vocabulary from their vast databases. Educational gaming, facilitated by tools like Secret Prompter, further stimulates creativity, demonstrating that the process of learning AI collaboration can be engaging and non-threatening.

The tension between AI-driven efficiency and the preservation of critical thinking remained a central theme in faculty discussions. Research within Ukraine indicates that while 84% of medical students use AI, 36% still view it as a form of academic misconduct, and over half admit to cheating. The IFNMU model addresses this by discouraging the prohibition of AI; instead, faculty are actively experimenting with assessment designs that require students to demonstrate understanding, such as submitting AI interaction transcripts or correcting AI-generated clinical errors.

Ensuring factual accuracy in high-stakes medical contexts requires particular vigilance. The program emphasized that AI should be viewed as a «cognitive instrument» requiring human verification. This «human-in-the-loop» principle is codified in the IFNMU policy, which prohibits the total replacement of teaching expertise with automated content and mandates human control over key administrative and clinical decisions. This is reflected in the institutional deployment of two custom AI assistants, which recorded over 3.200 interactions with a 4.3 out of 5 satisfaction rating. These assistants utilize university-specific knowledge bases, ensuring that responses regarding admissions and study planning are grounded in verifiable institutional data.

The development of this program occurred against the backdrop of the ongoing conflict in Ukraine since 2022. Paradoxically, these crisis conditions may have accelerated innovation. Faculty and administrators, already adapted to the major disruptions caused by the COVID-19 pandemic and subsequent military aggression, demonstrated a reduced resistance to radical change. The existential stakes of the conflict have made the value of educational transformation more apparent, as digital competencies are seen as vital for the international competitiveness and future resilience of Ukrainian medical graduates.

This study documents the successful implementation of Ukraine’s first comprehensive program to train medical faculty in AI technologies. Significant institutional engagement, substantial resource development, and active integration of AI into teaching practices demonstrate the program’s effectiveness. Key success factors included a multi-level program structure that accommodated diverse needs, strong institutional leadership commitment, the creation of a supportive community of practice, an emphasis on practical applications and quick wins, and ongoing support rather than one-time training.

Persistent challenges include the rapid pace of technological change, which requires continuous learning; balancing AI efficiency with the preservation of critical thinking; ensuring the factual accuracy of AI outputs in medical contexts; and adapting generic AI tools to specialized medical applications. These challenges appear inherent to educational AI integration and require ongoing attention rather than one-time solutions.

The program demonstrates the feasibility of large-scale faculty AI literacy initiatives even in challenging contexts characterized by resource constraints and broader societal disruptions. Our experience offers a replicable model for other medical education institutions. The high scalability potential suggests broad applicability across Ukrainian and international contexts.

Investment in faculty digital competencies represents not merely the adoption of innovative technologies, but a fundamental transformation of medical education for the twenty-first century. As AI capabilities continue advancing rapidly, the gap between AI-literate and AI-illiterate faculty will widen, with significant implications for student preparation and institutional competitiveness. Systematic faculty development programs, like ours, offer a pathway to ensure that medical educators can effectively guide students in developing the critical competencies necessary for AI-augmented healthcare practice.

Future work should focus on the long-term sustainability of faculty AI competencies, the impact on student learning outcomes, and the development of practical strategies for ongoing adaptation to rapidly evolving technology. Collaboration among institutions and sharing of effective practices will accelerate progress toward AI-integrated medical education that maintains the highest standards of academic rigor and ethical responsibility.