Patterns of human – AI interaction in Collaborative Decision Support Systems: A systematic review and cross-domain applications: Πρότυπα αλληλεπίδρασης ανθρώπου – ΤΝ σε Συνεργατικά Συστήματα Υποστήριξης Αποφάσεων: Μια συστηματική ανασκόπηση και διεπιστημονικές εφαρμογές

Abstract

Artificial intelligence (AI) is rapidly being integrated into collaborative decision support systems (CDSS), transforming decision-making in sectors such as health, education, industry, and public administration. The aim of this paper is to systematically review research on human–AI interaction in CDSS, following the PRISMA 2020 methodology. A total of 137 publications were examined and, after thematic analysis, classified into five fundamental collaboration patterns: (a) advisory/consultative, (b) co-creation, (c) dialogical/iterative, (d) multi-agent/group-based, and (e) autonomous/supervised. For each category, illustrative applications are described, along with challenges related to trust, transparency, and explainability. The findings indicate that CDSSs are evolving into dynamic, collaborative environments in which humans and AI assume fluid, complementary roles, underscoring the importance of active user participation and the provision of understandable explanations to enhance acceptance and credibility.