In this article for New Scientist, Claes de Vreese discussed findings of recent research around voters having political discussions with AI chatbots, and how the models actually have substantial persuasive potential. The applicability of the findings beyond experimental conditions, however, is yet to be explored.
Read more here:
