Framing the Human-Centered Artificial Intelligence concepts and methods: a scoping review. (Preprint)

This is the authors' abstract. We don't add key points for this paper.

BACKGROUND With the rapid expansion of Artificial Intelligence (AI) applications, researchers have begun focusing on the concept of Human-Centered Artificial Intelligence (HCAI). This field is dedicated to designing AI systems that augment and improve human abilities, rather than substituting them. OBJECTIVE The objective of the paper is to review the information on design principles, techniques, applications, methods and outcomes adopted in the field of HCAI, in order to provide some insights on the discipline, in relation with the broader concepts of Human-Centered Design and User-centered design. METHODS Following the PRISMA Checklist Extension guidelines, we conducted a systematic review in PubMed, Sciencedirect and IEEE Xplore, including all study types, excluding scoping review and editorials. RESULTS Out of the 1035 studies retrieved, 14 studies conducted between 2018 and 2023 met the inclusion criteria. The main fields of application were the health sector and artificial intelligence applications. Human-centred design methodologies were adopted in 3 studies, personas in 2 studies, while the remaining methodologies were adopted in individual studies. CONCLUSIONS Human-Centered Artificial Intelligence (HCAI) emphasizes designing AI systems that prioritize human needs, satisfaction, and trustworthiness, but current principles and guidelines are often vague and difficult to implement. The review highlights the importance of involving users early in the development process to enhance trust, especially in fields like healthcare, but notes that there is a lack of standardized HCAI methodologies and limited practical applications adhering to these principles. CLINICALTRIAL N/A