Abstract

Generative artificial intelligence systems are rapidly becoming embedded in the conversational environments through which organisations coordinate work. Large language models now draft emails, formulate recommendations, summarise discussions, and generate responses within collaborative platforms and messaging systems. While these systems produce linguistically coherent text, the organisational meaning of such messages cannot be determined by language alone. In rganisational settings, statements such as requests, proposals, approvals, and commitments derive their significance from the institutional roles of actors and the communicative infrastructures through which such acts are recognised. This paper argues that the emergence of generative AI reveals a structural limitation in the traditional information-processing view of information systems. When machines can generate language that resembles organisational communication, conversational environments risk blurring the boundary between linguistic expression and accountable organisational action. The organisational challenge of generative AI is therefore not only a problem of model behaviour but a problem of communicative infrastructure design. Building on the Language Action Perspective, actability research, and information infrastructure theory, the paper develops a design theory for accountable AI-mediated organisational communication. The theory conceptualises organisational coordination as a process of commitment formation occurring through communicative acts performed by actors occupying recognised roles. It then introduces the concept of accountability architectures—minimal infrastructural mechanisms that anchor conversational exchanges to  nstitutional roles, recognised communicative act types, and traceable acceptance of commitments. The design theory specifies core constructs, design principles, and propositions that enable conversational systems incorporating generative AI to preserve the institutional traceability of organisational commitments. The feasibility of the proposed approach is illustrated through the example of a minimal accountability wrapper for AI-mediated communication. By reframing organisational AI governance as a problem of communicative infrastructure design, the paper integrates several streams of Information Systems research and proposes a new design perspective for conversational systems supporting organisational coordination in environments where machines participate in the production of language.

Download

Share

COinS