Izvorni znanstveni članak
https://doi.org/10.31306/s.68.3.2
Beyond the algorithm: Thinking critically in occupational health
Lovela Machala Poplašen
; University of Zagreb, School of Medicine, Andrija Štampar School of Public Health, Zagreb, Croatia
Hana Brborović
; University of Zagreb, School of Medicine, Andrija Štampar School of Public Health, Zagreb, Croatia
*
* Dopisni autor.
Sažetak
Artificial Intelligence (AI) technologies are increasingly integrated into occupational health medicine practice, offering potential improvements in diagnostic accuracy and efficiency. However, their implementation introduces novel challenges, including algorithmic bias, lack of transparency in decision-making processes, and risks of over-reliance on automated outputs without appropriate clinical judgment. As occupational health decisions directly affect both individual worker protection and broader workplace public health, professionals must develop advanced information literacy that extends beyond traditional source evaluation to include a critical understanding of AI systems themselves, including their underlying data sources, technical limitations, and potential biases. This study evaluated AI information literacy among 16 occupational health professionals (specialists and residents) participating in the EASOM Summer School 2025. A five-item knowledge-based assessment adapted to occupational health contexts evaluated participants' competencies in identifying biased research questions, selecting evidence retrieval strategies, assessing the accuracy of AI-generated content, applying ethical AI use practices, and understanding proper citation protocols. Overall mean quiz performance was 2.56/5 points (SD = 1.09), indicating substantial gaps in AI information literacy. Questions assessing biased research design yielded the poorest performance (12.5% correct), while questions on ethical AI use in academia (81.2% correct) and citation practices (68.8% correct) demonstrated higher competency. Age showed significant negative correlation with quiz scores (ρ = −0.511, p = 0.043), suggesting younger professionals are better prepared for AI-integrated practice. These findings, although comprised from a small sample, demonstrate the need for targeted educational interventions addressing algorithmic bias recognition, critical question formulation, and responsible AI integration in occupational health medicine practice and worker protection.
Ključne riječi
Hrčak ID:
351747
URI
Datum izdavanja:
4.10.2026.
Posjeta: 0 *