Embedding responsibility in intelligent systems: from AI ethics to responsible AI ecosystems

Abstract Intelligent systems that are capable of making autonomous decisions based on input from their environment have great potential to do good, but they also raise significant social and ethical concerns. The discourse on ethics and artificial intelligence (AI) has covered these concerns in depth and developed an array of possible ways of addressing them. This article argues that a shortcoming of this discourse is that it concentrates on specific issues and their mitigation but neglects the nature of intelligent systems as socio-technical systems of systems that are often described as ecosystems. Building on the discussion of ethics and AI, the article suggests that it would be beneficial to come to an understanding of what would constitute responsible AI ecosystems. By introducing the concept of meta-responsibility or higher-level responsibility, the article proposes characteristics that an ecosystem would have to fulfil, in order to be considered a responsible ecosystem. This perspective is theoretically interesting because it extends the current AI ethics discourse. It furthermore offers a novel perspective for researchers and developers of intelligent system and helps them reflect on the way they relate to ethical issues.