Slim omgaan met AI in de zorg

Artificial intelligence (AI) offers opportunities for more efficient care, improved communication, and support in diagnostics and clinical decision-making. The transition to safe application in practice, however, requires AI literacy: the ability to understand the functioning, limitations, and risks of systems and to use them critically. The European Regulation on artificial intelligence (AI Act) obliges providers, deployers, and users to ensure an appropriate level of AI literacy. This article outlines the core aspects of AI literacy, with specific attention to predictive models and generative AI, including their pitfalls such as bias, confounding, hallucinations, and automation bias. The distribution of responsibilities is also addressed: providers are expected to ensure transparency and validation, deployers to establish governance and training, and healthcare professionals to apply contextual interpretation. Education and assessment of both basic knowledge and application are essential, from initial training to specialist practice. Only through early and structural investment in AI literacy can human oversight be safeguarded and patient safety maintained.

Conflict of interest and financial support: ICMJE forms provided by the authors are available online along with the full text of this article.