Exploring trust in a conversational news recommender through perceived enjoyment and personalisation

Conversational Agents (CAs) are transforming how news is distributed, offering a more dialogic and personalised approach for users. This transformation may help news organisations reach new audiences and hold the potential not only for them to engage with a new public, but also to enhance trust in their news recommenders. Despite the growing use of CAs for news recommendations, there is a limited understanding of how CAs in comparison to websites affect trust in the news recommender. To fill this gap, we assess the influence of various potential antecedents of trust (i.e. perceived ease of use, enjoyment, personalisation, and privacy concerns) through an online experiment. The experiment (N = 595) compared a news recommender in either a traditional news website or a CA. The results showed that participants had lower levels of trust in CAs compared to websites mediated by perceived enjoyment. This finding suggests that the dialogic nature of CAs does not inherently enhance enjoyment in the news context. Additionally, perceived personalisation was found to be an important antecedent influencing trust in the news recommender. Overall, these findings suggest that news organisations offering conversational news should prioritise enjoyable interactions with relevant content for the users to enhance trust.