Journalists in the loop: editorial governance of generative artificial intelligence

This article discusses the participation of journalists in the development and use of generative artificial intelligence systems. From a literature review, it proposes to understand the journalist as a developer of technological solutions, without reducing him to the function of programmer or end user. As an alternative to efficiency-centered automation, it articulates the human-in-the-loop principle to the editorial governance of AI, organizing human participation into three dimensions: (a) learning, (b) decision and operation, and (c) auditing. It argues that journalistic control should encompass the design, development, validation and continuous review of systems.