Toward an Artificial Deliberation? On Google DeepMind’s Habermas Machine

Google DeepMind has developed a system called the “Habermas Machine” in honour of Jürgen Habermas and his theory of deliberation. The system seeks to facilitate deliberation and assist individuals in finding common ground on controversial issues. This system, based on large language models, has been tested on groups in the United Kingdom, engaging more than 5,700 participants divided into groups. The study reported that participants preferred the AI-mediated group statements, considering them clearer, more informative and less biased than those generated by human facilitators. In light of Habermas’s theory of discourse and theory of deliberation in general, this paper aims to briefly raise some constructive remarks on this machine developed by DeepMind. Artificial intelligence can effectively be beneficial for deliberation. However, an artificial intelligence generator of consensus might bring about some theoretical and practical issues, concerning the very nature of the agreement, the moral inclusivity and, more broadly, its implicit rationality. Rather than being just theoretical, these issues involve the concrete future relationship between practical autonomy and artificial intelligence.