Online Deliberation with the Neutral, Unbiased and Artificial Facilitator

Limitations of human cognition invite skepticism to ordinary citizens’ capacity for reliable political decision-making. Deliberative democracy challenges this skepticism. There is a cognitive potential in deliberation (Habermas in Min, 2016) which addresses biases that undermine the quality of active listening and evaluation of others’ standpoints in deliberation. This occurs under designing and meeting certain normative conditions conducive to deliberation. In some approaches to these conditions, neutrality and impartiality are valued and expected of some participants. In online deliberation, solutions to problems of face-to-face deliberation are often met with a technical fix. With expectations of deliberation facilitators’ neutrality, an artificial intelligence facilitator is introduced to optimize the conditions for deliberation. As the normative conditions for deliberation have a cognitive potential, how important is neutrality in democratic deliberation? This paper uses the case of employing artificial intelligence facilitator, Sophie (in Wyss, 2018; Wyss & Beste, 2017) in online deliberation to uncover the import of neutrality as a deliberative norm. The neutrality of this artificial facilitator is employed for enhancing argumentative exchange and address motivated reasoning widely observed in online interactions (Wyss & Beste, 2017). We argue that neutrality is unattainable neither by human nor by artificial facilitators. To examine deliberation with the neutral, unbiased and artificial facilitator, we first explore theorization of neutrality and impartiality in institutionally designed citizen deliberation. Second, we reflect on the cognitive approaches that explain the cognitive transformation in deliberation, namely Kahneman's (2011) fast and slow thinking. Third, we highlight an alternative cognitive explanation based on the social function of reason (Sperber & Mercier, 2018) explains the cognitive potential of deliberation and how it tempers bias. Fourth, we propose some considerations for future uses and experimentations of artificial facilitators based on revisiting the import of neutrality in deliberation.