Human–AI interaction in knowledge ecosystems: a context–mechanism–outcome perspective

Purpose As human–AI interaction (HAI) becomes increasingly integrated into (inter)organizational knowledge practices, its role in shaping knowledge ecosystems (KE) remains insufficiently understood and calls for more contextualized inquiry and structured research agendas. This study aims to systematically review existing research that addresses HAI within KE contexts and situates it within the broader ecosystem literature by comparing KE with innovation, business and entrepreneurial ecosystems. This comparison identifies key overlaps, divergences and theoretical blind spots. Design/methodology/approach To examine how HAI contributes to achieving KE goals, the authors conducted a realist literature review. To ensure methodological rigor, the authors adopted the preferred reporting items for systematic reviews and meta-analyses) protocol. Findings The authors identify five key roles of HAI in KEs – assisting, conversing, learning, coordinating and decision-making – that support knowledge creation, development, transfer, boundary spanning and governance. These findings are synthesized into a context–mechanism–outcome model, which conceptualizes how different types of HAI (mechanisms) within KE (context) enable KE-related goals (outcomes). Originality/value To the best of the authors’ knowledge, this study offers the first systematic model of HAI mechanisms in KEs and links AI-related discourse across adjacent ecosystem types. The authors also provide contextualized examples of HAI applications and propose research questions that serve as a launchpad for academic–practitioner dialogue on the responsible and ethical integration of AI in KEs.