In a pioneering effort to bridge cultural nuances with technological advancement, a team of Japanese and British researchers has successfully developed and validated a Japanese version of the 12-item Attitudes Towards Artificial Intelligence scale (ATTARI-12), tailored specifically for medical students and resident physicians. This groundbreaking work addresses a critical gap in medical education in Japan, providing a psychometrically robust instrument to evaluate how emerging medical professionals perceive and interact with AI technologies. As artificial intelligence increasingly influences diagnostic precision, therapeutic decisions, and the landscape of medical training, understanding the clinician’s mindset is essential to ensure responsible and effective integration.
The original ATTARI-12 scale, introduced by Stein et al. in 2024, was constructed to capture the multifaceted nature of attitudes towards AI, spanning affective, cognitive, and behavioral dimensions. However, its direct application in Japan was hindered by the absence of a culturally sensitive and linguistically adapted version. The importance of this adaptation cannot be overstated, given Japan’s unique cultural context that includes elements of uncertainty avoidance and distinct social norms influencing technology acceptance. The newly validated Japanese adaptation, the J-ATTARI-12, thus represents not only a translation but a cross-cultural transformation, making it a vital tool in the era of AI-driven medicine.
This translational research was spearheaded by Project Assistant Professor Hirohisa Fujikawa from Juntendo University Faculty of Medicine, who collaborated with colleagues from both Japanese institutions and Durham University in the United Kingdom. Their work followed rigorous international consensus guidelines for translation and cultural adaptation. The process involved exhaustive forward and backward translation, expert review panels, and pretesting among medical trainees to ensure conceptual alignment and contextual relevance. The research team then conducted a comprehensive nationwide survey in mid-2025, recruiting 326 participants spanning multiple medical schools and teaching hospitals. This sizeable and heterogeneous sample provided a robust foundation for psychometric evaluation.
To establish the reliability and factorial validity of the J-ATTARI-12, an exploratory factor analysis (EFA) was performed on half of the sample, revealing a compelling two-factor structure reflecting diametrically opposed dimensions: “AI anxiety and aversion” contrasted against “AI optimism and acceptance.” This bifactor model aligns with contemporary theories of technology acceptance, which acknowledge both apprehensions about ethical responsibility and potential job displacement on one hand, and enthusiasm for AI’s role in augmenting clinical decision-making and personalized education on the other. Confirmatory factor analysis (CFA) on the second half of the sample established that the two-factor model significantly outperformed a one-dimensional structure, demonstrating excellent model fit indices.
A critical component of validating the J-ATTARI-12 involved assessing convergent validity. This was achieved by correlating scores from the scale with measures of attitudes toward robots—a complementary and culturally resonant technology. The moderate positive correlation discovered underscores the construct validity of the scale, indicating that medical trainees who generally exhibit positive attitudes toward robotic technologies likewise tend to embrace AI applications. Internal consistency reliability, quantified by Cronbach’s alpha, showed high values for both subscales and the total score, affirming that the instrument reliably captures nuanced attitudes without internal redundancy.
The implications of this research extend well beyond psychometric properties. Medical educators now have access to a validated tool capable of identifying learners’ uncertainties, hesitations, or biases related to AI incorporation in clinical practice and training curricula. This capability is vital, as Dr. Fujikawa emphasizes, because attitudes significantly influence whether AI will be adopted responsibly and effectively. With ongoing AI technological advancement, medical training programs must remain responsive to learners’ evolving perspectives to design targeted interventions, fostering openness while addressing ethical, privacy, and autonomy concerns that persist among trainees.
Further, the J-ATTARI-12 enables longitudinal tracking of attitude changes as AI becomes more embedded within healthcare systems. Such data provide empirical grounding for curriculum development and policy-making, ensuring that AI education keeps pace with technological innovation while safeguarding humanistic values in medicine. Notably, Juntendo University plans to integrate this scale into its new “Medicine and AI” program set for launch in 2026, which aims to cultivate a generation of clinicians proficient in AI-enabled practice. This foresight underscores the importance of coupling technological proficiency with attitudinal readiness in medical education frameworks.
The broader societal relevance of this study touches on Japan’s historical relationship with medical education and innovation. Founded in 1838 as a Dutch School of Medicine, Juntendo University has long been a pioneer in integrating Western medical advancements into Japanese healthcare. The institution’s commitment to continuous evolution, as encapsulated in the motto “Jin – I exist as you exist” and the principle of “Fudan Zenshin – Continuously Moving Forward,” is now reflected in embracing AI technologies with cautious optimism. The J-ATTARI-12 exemplifies this ethos by ensuring that human considerations remain central amidst rapid technological change.
Technical sophistication aside, this study also highlights crucial cultural dynamics affecting AI uptake in medicine. The Japanese context, characterized by collective social values and high uncertainty avoidance, necessitates approaches that are sensitive to emotional and normative dimensions of technology acceptance. The identification of “AI anxiety and aversion” as a distinct factor signals areas where educators and policymakers must engage learners empathetically to alleviate fears while promoting informed optimism. Such insights deepen the global understanding of how culture shapes AI integration in healthcare.
Moreover, this research sets a precedent for cross-national collaborations in health professions education research. By partnering with UK-based scholars, the Japanese team incorporated diverse perspectives and methodological rigor, enriching the adaptation process. This strategy not only enhances the validity of the J-ATTARI-12 but also opens pathways for comparative studies across different countries, fostering a global dialogue on AI attitudes among future healthcare providers. Such efforts are crucial as AI transcends borders, requiring harmonized yet contextually informed educational strategies.
The methodology of this study exemplifies best practices in scale adaptation research, including the use of large multicenter samples, split-half validation approaches, and multiple convergent validity checks. This robustness responds to prior limitations in attitude measurement, where linguistic inaccuracies or cultural mismatches weakened scale reliability. By addressing these methodological challenges, the J-ATTARI-12 offers a blueprint for other researchers seeking to translate AI attitude scales into diverse languages and settings, advancing the field of medical education research internationally.
Finally, the J-ATTARI-12 scale stands as a critical link between technological advancement and human factors in healthcare, emphasizing that successful implementation of AI depends as much on users’ acceptance as technological performance. As Dr. Fujikawa poignantly notes, making these attitudes visible allows for better education and more responsible AI adoption. This tool thus empowers medical educators, researchers, and health systems alike to navigate the complexities of AI integration, ultimately enhancing patient care and preserving the humanistic spirit of medicine.
Subject of Research: People
Article Title: Adaptation of the Japanese Version of the 12-Item Attitudes Towards Artificial Intelligence Scale for Medical Trainees: Multicenter Development and Validation Study
News Publication Date: 14-Jan-2026
Web References: http://dx.doi.org/10.2196/81986
References: DOI: 10.2196/81986
Image Credits: Hirohisa Fujikawa from Juntendo University Faculty of Medicine, Japan
Keywords: Artificial intelligence, Computer science, Applied sciences and engineering, Health and medicine, Human health, Health care, Scientific approaches, Scientific community
